PC Build for Photoshop CS2 / Best Bang for the Buck

RT
Posted By
Randy_Triplett
Apr 11, 2007
Views
2249
Replies
128
Status
Closed
I want to build a new Strictly for Photoshop CS2. I will probably run XP for a while but may switch to Vista eventually.

I prefer my pc’s to be as silent as possible. I have a power supply and case in mind, full ATX, and I have a 7600GT fanless card that I will probably use.

What processor and memory combination will be most efficient for Photoshop?

How to Improve Photoshop Performance

Learn how to optimize Photoshop for maximum speed, troubleshoot common issues, and keep your projects organized so that you can work faster than ever before!

DP
Daryl_Pritchard
Apr 12, 2007
Randy, I don’t know what CPU/memory combo is the most efficient, but you will get very good performance from an Intel Core 2 Duo processor and 2GB of memory, while more memory is better still. Many threads in this forum discuss PCs and what is best, so I suggest you just do a bit of searching and review the many opinions shared. There are of course different Core 2 Duo processors and if you don’t plan to overclock, then I’d suggest you simply buy the fastest CPU you can afford. Quads are out now also, but I’m not sure of their prices and whether they are cost-effective yet. As for myself, my system runs PS CS2 well but also wasn’t exactly cost-effective. I bought the Core 2 Exreme processor and 4GB of DDR2-1066 Corsair memory that hasn’t performed at the levels I expected, but that appears more due to motherboard limitations for my ASUS P5W-DH Deluxe with. I’m not complaining though…I’m running at 3.6GHz with high stability and have only seen one odd and minor PS behavior that I cannot identify a cause for at the moment, but the overclocking could be a factor.

I too have a 7600GT card and it works well, but I might also reconsider getting an ATI card if I were do rethink this system. Most, if not all, ATI cards have true dual LUT support which makes it even easier to run two monitors with independent color profiles, should it be you plan on using two monitors. With nVidia cards however, there is still a workaround, using the Windows Color control panel utility or perhaps some 3rd-party gamma loaders.

My system, if you’re interested, is at <http://ambress.com/pc>

Regards,

Daryl
TT
Toby_Thain
Apr 12, 2007
Macbook/Macpro/iMac
BL
Bob Levine
Apr 12, 2007
Yup. They run Windows quite nicely now that Apple is using real processors in them.

Bob
RT
Randy_Triplett
Apr 12, 2007
Thanks Daryl,
Thanks for your comments. You’ve spent a little more than I have budgeted, so I think I will probably keep in the Core 2 Duo set, probably the E6600, which is currently in the $300+ range and keeps up pretty well with most except the "extreme" processors. I’m supposing that I will get at least 2gb memory and will shoot for 4gb.

I’m also interested in a really good socket 775 mobo. Any suggestions for that? (Daryl, you seem to be saying you’re not totally satisfied with your Asus board…)

Thanks again,
Randy
RT
Randy_Triplett
Apr 12, 2007
Toby and Bob, What part of the topic title "PC Build for Photoshop CS2…" do you not understand?

If you must know, I promised my nephew I’d show him how to build a pc this summer, if you don’t mind.

Thanks anyway, but your "suggestions" don’t exactly address my question.
BL
Bob Levine
Apr 12, 2007
Randy,

I think you totally misunderstood my post. I was pointing out in a very sarcastic way just what you said.

Building your own PC can be a very rewarding experience. Nothing like taking a pile of electronics, putting it all together and turning it on for the first time.

Have fun with it.

Bob
RT
Randy_Triplett
Apr 12, 2007
Sorry Bob,

I was kind of taken aback by Toby’s response, and didn’t take time to recognize your sarcasm.

You could make use of the winky face 😉 lol

Randy
JZ
Joe_Zydeco
Apr 12, 2007
Toby and Bob, I think you guys were advising Randy on "Best Bang for Really BIG Bucks!" 🙂
BL
Bob Levine
Apr 12, 2007
I’m sure there are people buying Macs now to run Windows. These are high end machines and if you compare them to some of the higher end machines out there, the price really isn’t out of line.

You can’t even buy an Inspiron from Dell with the processors that Apple uses in the Mac Book Pros.

Bob
JZ
Joe_Zydeco
Apr 12, 2007
At whatever performance level, a mac always costs more than a PC because of Apple’s monopoly on the hardware. With no competition, they can price the things wherever they like. Note that Intel processors were much higher priced until AMD became a major competitor. I support AMD not only because it is a good company, but because if AMD were to go under, Intel would run their prices right back up into the stratosphere again.
DP
Daryl_Pritchard
Apr 12, 2007
Randy,

Overall, I do think ASUS makes good motherboards and they generally get very good reviews from what I’ve seen, so don’t let my comments steer you away from them. My board seems to be solid, but I just haven’t been able to overclock it with the particularl memory timings I was aiming for. That could be due to the board, the memory, or both. Namely, while only spec’ed for supporting DDR2-800 memory, the BIOS allows for selecting a DDR2-1066 memory rate if your memory is up to the task as I thought my Corsair DDR2-1066 memory would be. That wasn’t the case however, and Corsair even reports they’ve only gotten that memory to run successfully at DDR2-1066 speeds using, I believe, AMD chipset motherboards. DDR2-800 performance is the best they obtained with this ASUS P5W-DH motherboard. I’ve not even obtained that from it, but perhaps could if I ran with looser CAS timings. As it is, I’m running at DDR2-730 with tighter timings that give me a combination of performance and stability that so far seems to be optimum for the components I’ve got.

I’m happy with the 3.6 GHz operation the Core 2 Extreme is delivering without having to go for too fancy a cooling solution. Still, I’ve read others obtaining near the same level of performance with less expensive Core 2 Duo processors and memory, so there’s definitely some money to be saved over what I spent.

Regards,

Daryl
RT
Randy_Triplett
Apr 12, 2007
Now, kids, I’m not taking sides, (although I have strong opinions about this debate,) I’m just trying to build a pc.

Let’s not get too far off topic 😉
RT
Randy_Triplett
Apr 12, 2007
Daryl,

You are apparently more into overclocking than I am. For my use, I’d almost rather underclock my systems to get cool running and low fan noise, or no noise at all, which I’ve achieved from several builds.

Good information though. Between you and Tom’s Hardware, I’m getting pretty close to having my specs in line…

Thanks
Randy
DM
dave_milbut
Apr 15, 2007
now that Apple is using real processors in them.

zing! 🙂

I think RL’s been lurking in the lounge! 🙂
BL
Bob Levine
Apr 15, 2007
Not a chance, Dave. 🙂

Bob
DP
Daryl_Pritchard
Apr 15, 2007
Randy,

This is my first system ever to overclock, but I did buy the components to facilitate that…or so I thought. As I said earlier though, I’ve since learned that even the lesser Core 2 Duo CPUs overclock quite nicely and with little need for extravagant cooling solutions. True enough, my system isn’t silent, but it sits on the floor and is just a soft hum that is perhaps the quietest of any system I’ve had in the past. The blower on my memory that I added, as well as the front fan, are both controllable, so I can easily run them at quiet levels while still getting good airflow in conjunction with other fans in the system. My temps are typically in the 34-37°C range unless benchmarking, when I’ve seen the CPU temp peak up to about 43°C.

Daryl
P
PECourtejoie
Apr 16, 2007
Randy, I guess that you know about silentpcreview.com ?
A good, silent and efficient PSU is the first thing I’d get.

Get several HD’s: system+programs, Scratch disk, OS pagefile, and external drive for backup, max the ram (4GB), maybe use a 64bits OS if all your periphericals support it (well, if they have drivers)

As it was said Intel Core 2 duo processors are the best now.
RT
Randy_Triplett
Apr 16, 2007
I usually just use two large fast drives internally and partition the second so the first partition of about 10 gb is dedicated to the Windows page file and the scratch disk, and the rest for data. I also use several backup schemes and I’m staying 32 bit for now.

Is there a problem with the scratch disk and the page file being on the same partition?

And yes, I use silentpcreview.com and <http://www.endpcnoise.com/> and <http://www.quietpcusa.com/> are also good sources.

Thanks
LH
Lawrence_Hudetz
Apr 16, 2007
I believe that the principal reason for the scratch to be on it’s own drive is to separate scratch from paging.I have paging separate from "C" and scratch, each on it’s own drive.
DP
Daryl_Pritchard
Apr 16, 2007
Should it be that you go with 4GB of memory, I believe it is usually recommended that your pagefile be 1.5x your memory, or 6GB. How true that is, I’m not sure, as it may be a rule based in the day where having so much memory was uncommon. Regardless, should you go with that approach, be aware that any given pagefile can be no larger than 4095MB. I think that’s a 32-bit O/S limitation. So, anything more than that will require your pagefile to be distributed to multiple drives. I allocate a fixed 4095MB to a separate drive, while another fixed 2048MB resides on my system drive. And, similar to Lawrence, my PS Scratch disk is on yet a 3rd drive.

Regards,

Daryl
RT
Randy_Triplett
Apr 16, 2007
Thanks Guys,

From my experience and from everything I’ve been able to read, although 1.5x the ram for the page file used to be recommended, I find that running a single same size page file, whether it’s 2gb or 4gb is pretty much ideal. Maybe it’s just my system, but I get the best speed running that way.

But you guys actually have separate physical drives for the PF and the scratch? I would have thought that separate partitions would be adequate. I always put my pf on a small partition at the beginning of my second sata drive.

What you suggest makes some sense to me, but are you using older small drives? If you are, aren’t newer drives faster? Seems to me as long as the OS and PS are on the primary and page and scratch are on the secondary, that would be enough…

No?
Randy
DM
dave_milbut
Apr 16, 2007
I would have thought that separate partitions would be adequate.

nope. that still means the heads have to keep going back and forth on the same drive. the point is to keep the files seperate so that on can read and write the systme file while one can read and write the ps swap file.
RT
Randy_Triplett
Apr 16, 2007
Dave,

That makes sense, but I have my applications and PS files on my primary and the first partition of the secondary houses the page file only. The second partition of the secondary drive is inactive when I’m using PS.

But in either case, I can see your point and it would be no trouble to dig up a couple of old small drives. As I am concerned about noise, I was just concerned about the extra heat from extra drives…

Randy
DP
Daryl_Pritchard
Apr 16, 2007
Randy,

Thanks for the feedback on the pagefile size. I’ve never attempted to measure any sort of performance hit or gain from changing the pagefile size, and I somewhat thought a 4GB size should be more than enough for most activities. In fact, as PS is my most demanding application to run and yet uses its own scratch disk (although I don’t recall if this is in lieu of or in supplement to the pagefile), I’d not expect the pagefile would need to be very big.

As for all the "where to place" issues regarding the pagefile, scratch disk, etc., keep in mind too that this is all theoretical. Sometimes I wonder how much real difference one would observe if both pagefile and scratch disk were on the same partition or same drive. Regardless, I would suspect the less thrashing around that hard drive heads have to do, the less risk there is of data corruption or disk error.

I’ve jumped into Windows Vista this weekend, setting it up for dual-boot operation so I can explore it for an upgrade down the road. I like the looks of it somewhat, yet may still find favor with the classic interface. Aero will take some getting used to before I decide whether or not its an improvement. Meanwhile, it was frustrating to see how Microsoft has jacked with the security controls, with permissions on many things not changeable unless you take ownership of them as Administrator/Administrators. I need to study up on the ramifications of doing that. They’ve also mucked around and moved certain directories to new locations, dummying-up the system with shortcuts named per the old conventions, and pointing them to the new areas. I’ve no idea why they’ve taken this approach, as it just strikes me as much better organized to, as an example, have the All Users, Default User, and individual User files all in Documents and Settings. Ah well, just more to learn.

Daryl
RT
Randy_Triplett
Apr 16, 2007
Daryl,

Good for you for trying out Vista. I suppose we all will some day, although it seems poorly conceived to me, albeit it more secure and "prettier." My experiences with it have been few, but I was not impressed especially when asked to fork over so much money for it. Frankly, I’m happy to run my necessary apps, like CS2 on XP and in the meantime I’m looking hard at Ubuntu for my everyday casual use. It’s not hard to put together a really interesting, and fast machine all running free software, including Beryl, which may just be "prettier" than Aero! I think Windows is going to alienate all enthusiasts who like to build their machines with this price structure and intrusive OS. They’re going too "Big Brother" for me to grin and bear it. But that’s just me.

On the pagefile issue, this article sums up what I found in my search:

<http://www.theeldergeek.com/sizing_the_page_file.htm>

After trying and benchmarking several setups, I found that with my 2gb of ram, a 2gb page file, (placed where I’ve already mentioned,)seemed to maximize my performance.

I’m certainly no expert, and I didn’t really do any very official looking benchmarking, but it works for me. I guess the key is to ask the computer to use the much faster ram rather than the page file, no matter where it is… YMMV

Good luck.

Randy
TS
Tom_Scudder
Apr 16, 2007
I just built a new computer and I would suggest a Raid5 system. I installed three 400gig HD’s. Much more secure in the fact that if one drive fails all I have to do is replace it and the system automatically adds the new drive into the Raid5 array….no loss of data. Great invention
TS
Tom_Scudder
Apr 16, 2007
I would suggest adding a Raid5 array to your system…Peace of mind in case of a single HD crash. I installed three 400gig HD’s into my system. If one crashes, just replace it and the system will restore to match the other two. Great invention for peace of mind
H
Ho
Apr 17, 2007
I know the conventional wisdom says to separate Page and Scratch files. But in my old computer I put them both on a single, unpartitioned SCSI ultra 160 drive (10,000 RPM) and didn’t see a performance hit, at least while I was benchmarking. Maybe in real life an occasional gain can be had from using discrete drives, but for the most part I don’t think it matters.
RT
Randy_Triplett
Apr 17, 2007
HO,

I tend to think you are right. With the speed of today’s processors and especially today’s ram, I think the page and scratch files are not even utilized that much. And when they are, the impact is minimal. I’ll go for more ram…
LH
Lawrence_Hudetz
Apr 17, 2007
Not much, until your files get big.
AJ
Adam_Jerugim
Apr 17, 2007
Some system tips…

CPU: Right now the most efficient processor on the market is the Core 2 Duo. The high-end AMD Athlon64 FX series CPUs are also very good, but they are more expensive (last I checked the only systems I have that support 2000MHz FSB are the newer Athlon/Opteron systems).

Hard Drives: We’ve seen problems with RAID5 scratch drives, so instead I would suggest the following…

* a very fast SATA drive for your primary OS volume (the volume you install and run PS on) * at least two very fast SATA drives striped as RAID0 dedicated as PS scratch

(I use multiple 74GB 10K RPM Raptor drives and have been very happy with the overall performance)

RAM: Unless you’re using Windows XP64 or Vista 64bit, you don’t need more than 4GB physical RAM. If you are using a 64bit Windows OS, then pack as much RAM as you can afford into the system (VM OS buffering kicks in for PS with over 4GB of RAM in 64bit systems and allows PS to access nearly all of your system RAM). The faster the RAM and the system’s FSB, the better.
DD
D_Davies
Apr 19, 2007
Agree mostly with Adam, either Core2duo or Quad (prices coming down end of May?). My scratch is a single Raptor and works well, and I multi-task in 32bit XP so 4gb is pretty much essential. Bear in mind Vista saps RAM so the more the merrier. Also bear in mind on these forums I have heard people saying that 64bit Vista is slow with PS3, 32bit is not. That’s the rumour.

– D
RT
Randy_Triplett
Apr 19, 2007
Thanks guys,

I am sticking with 32 bit for now and I have little interest in Vista. I also am only a serious amateur, so an array of large Raptors is pretty much out of the question too, just for economy’s sake.

I imagine going from a 2.6 ghz P4 with 2gb ram, to a Core 2 Duo E6600 with 4gb of ram, and a clean system without those 70+ services I have running now on my main system, will impress me sufficiently. I’m pretty sure I’ll notice a difference… 😉

Randy
RT
Randy_Triplett
Apr 19, 2007
BTW, I hadn’t heard about a big cpu price drop at the end of May, is that pretty solid?
KC
Kevin_Crow
Apr 26, 2007
The CPU price drop was April 22nd. Look at the e4300 and overclock to 3-3.2Ghz with good CPU Fan (Zalman 9700). If you are working on large files in PS then the memory bandwidth is as important as CPU speed, so try to max out your FSB (maybe lowering the multiplier on the CPU to get higher FSB hence higher memory. I assume you have paired memory banks for 2X bandwidth.). Photos are getting bigger and bigger because of the Megapixel growth in cameras today, so for photoshop on XP I have heard the 6GB is the sweet spot. PS will use memory over 3.5GB for scratch disk, before it uses HD scratch disk.

Kevin
RT
Randy_Triplett
Apr 26, 2007
Thanks Kevin,

I have seen the price drop in action and will be ordering parts soon. I’m still considering an E6600 but I’m not crazy about overclocking as I want to keep things as cool and quiet as possible.

I haven’t picked out a mobo or memory yet. Can you point me to any info about maximizing the FSB? I’ve never looked into this as I’m not much of a tweaker, but I’ll try anything once… 🙂

The thought of 6gb or more of ram just trips me out. I can remember when that was a nice sized hdd… 0_0

Randy
LH
Lawrence_Hudetz
Apr 26, 2007
Ditto on overclocking. Your particular board will have a part to play as well.

The E6600 is a good compromise with cost.

How in hell do you use 6G ram on a 32 bit system?
DM
dave_milbut
Apr 26, 2007
I’m still considering an E6600

I’ve got the 6600 with 2 gig ram on an intel board (975bx). no overclock. runs like a dream. fastest system i’ve ever (personally) seen. and i’ve seen a lot. i KNOW there are faster, but man this sucker flys!
DP
Daryl_Pritchard
Apr 26, 2007
Keep in mind that any amount of RAM over 4GB is wasted in a 32-bit O/S. Even 4G just gives you the ability (with WinXP and the 3GB boot.ini option) to allocate the upper 1GB for O/S use exclusively so the full 3GB below that is freed up for applications to use.

Randy, I don’t know much about FSB optimization, but I’ve read that the best performance is obtained from a 1:1 FSB:DDR ratio. In this case, FSB is not the "quad-pumped" value but rather the base FSB frequency, also called the CPU clock frequency. And as the DDR means "double date rate", then a DDR800 frequency means you’re actually using a 400MHz clock. So, as least with the AMI BIOs of my ASUS motherboard, if you specify a CPU frequency of 266 and a DDR2-533 memory frequency, that would provide a 1:1 ratio. Everything I’ve read about Core 2 Duo processors suggests they should be easily overclockable without requiring noisy cooling solutions if you keep the overclocking to a moderate level. I think I’ve got my system pushed as high as I can get it with nothing more than a fan-cooled heatsink design (Zalman CP9500), and I’m getting a 25% overclock on my 2.93GHz Core 2 Extreme, running it at 3.7GHz. If I split the difference at 3.3GHz, I’ll bet I could drop fan speeds enough that the noise factor would truly be minimal while still providing a performance boost to keep Photoshop happy.

Daryl
RT
Randy_Triplett
Apr 27, 2007
Thanks Guys, that’s a lot of good information. I’m glad I started this thread early so I had time to get all this good scoop b4 I order parts.

Daryl, you’re a wealth of info. I still don’t know if I’ll be overclocking, although these chips seem to be built for it. I really like quiet and stable. Emphasis on quiet… and on stable. We’ll see.

Dave, I’m looking at that setup, maybe with 4gb of ram, that or the Asus board. Still haven’t landed on the board/memory combo, but I’m getting there.

Lawrence…

" How in hell do you use 6G ram on a 32 bit system? "

LOL, I’m soooo glad you said that.

Randy
DM
dave_milbut
Apr 27, 2007
that or the Asus board.

imo, spend the couple extra $ and go intel. i’ve had nothing but luck with this one and the one before it (a d865perl with my p4 2.8c). that last one was the first intel board i ever had and i don’t see myself ever going back for features and compatibility issues. everything that’s SUPPOSED to work DOES work. every time.
DM
dave_milbut
Apr 27, 2007
oh and get GOOD name brand RAM at the max speed your board (whatever you get) supports, not something like "RAM-bo" RAM. <snicker>
DP
Daryl_Pritchard
Apr 27, 2007
Randy,

I have wondered if there might be a bad RAM slot on the board. I didn’t note which slot was in use by the failed modules at the time each of them failed. But, it could well have been the same slot. The good thing is that even if a pair of modules has to be replaced, I’ve still got 2GB keeping me pretty happy and running fine, but with just a bit less headroom.

For what it’s worth, I’m not encouraging you to overclock, but your question about memory optimization led me in that direction again. After all, a certain FSB:DDRAM ratio might be optimum yet the CPU multipliers might take you slightly below or slightly over the CPU spec frequency. In such cases, I’d definitely bump it up a notch rather than drop it any. And of course a faster bus operation alone should yield some performance gains even if the CPU frequency is maintained at stock levels.

Dave,

I thought…and pretty much still do…that Corsair was good name brand memory, yet one module has failed from each of two matched pairs I initially purchased. Fortunately, Corsair is good about replacing them so long as they’ve not be operated at higher-than-spec voltages (mine haven’t). Still, this also leaves me to wondering just how much more prone to failure high performance memory modules are. Of course, as I mentioned, there could be a RAM slot issue on my mobo…I sure hope not, as I’d hate to have to swap that out and truly have my system down for a while.

Daryl
LH
Lawrence_Hudetz
Apr 27, 2007
Daryl, if you are getting blue screened, you are at one or the other limit of your system margin, and as the temperature rises, you go over the limit. It doesn’t take much at that point.

I have to emphasise "system margin" here.

If you have a bad slot, the board is hosed, FAPP.
RT
Randy_Triplett
Apr 27, 2007
I have to say I have not seen a blue screen for about 4 years or so unless it was a hardware issue, and I currently have 8 computers running. Many are Asus boards. None are overclocked. All are made to run quiet via many methods. That’s the way I like it.

Occasionally over the maybe 15 builds and the 20 or so rebuilds, (for other people,) I’ve done in the past 3-5 years I’ve run into bad ram slots a couple of times. It usually isn’t a big deal as these systems are almost never maxed out and I’ve usually been able to get by on two slots without dumping the board. I never had a bad slot kill the ram though, at least as far as I know.

For my use, as I said I’m just an avid amateur, I don’t really need to maximize the performance of a new system, I was just curious. My builds are usually last gen high end that is now affordable but still smokes compared to my last build. I don’t need bleeding edge.
LH
Lawrence_Hudetz
Apr 27, 2007
Good moves, Randy. I contract for Intel (FSB) and my personal conclusions, seemingly valid for both brands, is set it to the spec and forget it. Your processor life will not be degraded that way.

Be sure to monitor the cpu temps.
DP
Daryl_Pritchard
Apr 27, 2007
Randy,

Likewise, I’m just an "avid amateur" as well, and this is the first system I’ve built where I anticipated overclocking it, largely due to how forgiving the Core 2 Duo family seems to be, especially in its thermal efficiency. I don’t need to run my system maxed out at 3.7GHz or whatever it "can" reach, and particularly not if it proves unstable with any number of BSODs. At 3.6GHz, I don’t recall ever seeing a blue screen but I’m certain 3.5GHz was "clean" in that regard. And 0.2 GHz isn’t such a gain as to merit sacrificing a pricey processor for. I’ve just been trying to learn what truly are the limits, given my component choices.

As for the bad memory, do either you are Lawrence know if it’s likely a board could have a memory slot that is just "on the verge" of being bad, such that over time it causes a module to fail, yet it otherwise appears to be working normally? I’m guessing all my slots are good, since I am generally running my system 24/7 without a hitch. As I bought the first 4 modules at the same time and yet one module of each pair went bad within 1 month of one month of each other, at about 8 months into the life of the system, it’s hard to know what to make of that. I don’t attribute the failures to overclocking since the memory was running well within its max. specs. I really wish I had kept notes on the first memory pair’s lot number, so I could’ve compared it to the second pair; I know the serial numbers were quite different, so I’m guessing they were of different lots, but I may be wrong. All I know to do now is wait on the new modules, install them, and see what the future holds for them in terms of lasting or failing.

Daryl
LH
Lawrence_Hudetz
Apr 27, 2007
never saw "on the verge" myself, but then, a bad slot is very rare, AFAIK. It will show up in BIOS, if it is populated. Your BIOS should beep a code, which the mobo book should be able to decode.

Since we deal in ones and zeros, no half way occurs! 😀

Life failures accrue to both metal migration and overheating when you overclock, as the overall current levels increase. So, it may not be excessivley hot, especially if you are running in a cold room, but still have failures.

Metal migration is rather slow, however.
DM
dave_milbut
Apr 27, 2007
could it be a spiking psu that’s frying chips intermittantly?
LH
Lawrence_Hudetz
Apr 27, 2007
Spikes are one element that figures into the overall scheme. Occasional spikes cannot be accounted for, but recurring ones do.
DP
Daryl_Pritchard
Apr 27, 2007
Thanks Lawrence,

What you’ve said is pretty much in line with what I was thinking, and my mobo has never beeped at me unless I deliberately coaxed it into doing. I usually keep the temps in my house around 72°F (22°C) and with the cooling I’ve got in my PC I rarely see the CPU go over 45°C under load, save for more strenuous benchmarks. The motherboard generally runs around 40-42°C.

Ah…well I did just get a peak of 51°C on my CPU temp: I ran a Radial Blur "benchmark" (Best Spin on 8×10-inch, 300ppi, 8-bit image filled with 400% uniform mononchromatic noise) in PS CS3 six times consecutively and, during the last run, launched Norton Antivirus and Spysweeper scans concurrently. When the final run of the Radial Blur ended, the CPU temp returned to 44°C. It was at 36°C before I began any of this. It was interesting to see the impact of the scans on PS performance; 26.2 seconds to complete the Radial Blur with PS CS3 alone running; 35 seconds with both scans added to the mix.

Dave,

I think my power supply is solid. Acutally, I’ve no reason to expect anything is a problem other than the memory itself, but that will just take time to assess. Once I get the 2 replacement modules, if I again see failures of the memory within the next 6-8 months, I’ll be more suspicious of something in my system being awry.

Daryl
RP
Russell_Proulx
Apr 27, 2007
I’ve got the 6600 with 2 gig ram on an intel board (975bx).

Dave,

Any reason why you chose the 975bx over a 965LT? Otherwise We have very similar systems. I’ve been using Intel processors on Intel motherboards for my last 2 systems and am *very* pleased with the stability and performance.

Russell
DM
dave_milbut
Apr 27, 2007
Any reason why you chose the 975bx over a 965LT?

<spinal tap>well it’s 10 better, ain’t it?</spinal tap> 🙂

no actually that was on of hte boards offered with my barebones config. and i just looked at what was the most expensive intel board and started comparing features (as opposed to specs). the diff between this board and the next one down (which may have been a 965) wasn’t that much, like 20 bucks, iirc, so thinking i’d want upgradability down the road, i decided to go with this one.

here’s the board:
<http://www.intel.com/products/motherboard/D975XBX/index.htm>

and yours:
<http://www.intel.com/products/motherboard/DP965LT/index.htm>
RT
Randy_Triplett
Apr 28, 2007
never saw "on the verge" myself, but then, a bad slot is very rare, AFAIK. It will show up in BIOS, if it is populated. Your BIOS should beep a code, which the mobo book should be able to decode.

Daryl, I’m with Lawerence on this one. It’s very uncommon and it works or it doesn’t, which leads me to think you have just had a run of bad luck on the memory.

I don’t often multitask the same machine as there are three pc’s under my desk doing different things, (thank you kvm,) and my processor never spikes over 45. They call me Captain Caution. Ta-Dumb!

Oh, and I can’t tell you how many problems I’ve traced back to power supplies over the years. Maybe the most important least respected component in the whole darned pc. Get a good one.

Randy
LH
Lawrence_Hudetz
Apr 28, 2007
Power supplies are not created equal. In my lab at Intel, we have a fellow who is constantly checking out power supplies, under load.

Buy the heaviest one for the rating from the best suppliers.
DP
Daryl_Pritchard
Apr 28, 2007
I did that…the Seasonic S12-600 p/w I bought is highly rated from all I’ve read about it and is spec’ed for 600W continuous, not peak, output. It is also among one of the more efficient power supplies, which I was also looking for.

I have always thought it funny that yes, power supplies are one product where heavier usually does mean better. In that case, mine is about 5 lbs.
RP
Russell_Proulx
Apr 29, 2007
so thinking i’d want upgradability down the road, i decided to go with this one.

Have you tried popping in 2 PCIe videocards (not as a Crossfire config) in the 975 board to see how they behave? I tried with this 965 but could not get it to work properly. No big deal since I usually use only one monitor but I tried just to see if it was possible.

Russell
GM
gordon_morris
Apr 30, 2007
Randy,

When you order your PC – gleaned from all the input at your disposal – would you mind posting the specs here?

thanks,

GM
DP
Daryl_Pritchard
Apr 30, 2007
Having joined the "me too" generation of CS3 users this weekend, I spent some time running a popular Radial Blur benchmark on PS CS2 and PS CS3 for both WinXP and WinVista. The benchmark involves timing the execution of a 100% Best Quality Spin Radial Blur on an 8×10-inch, 300ppi image filled with 400% uniform monochromatic noise. At the time, I was running my Core 2 Extreme system at 3.6GHz with 2GB of dual-channel memory installed and was very pleased to see PS CS3 delivering roughly a 40% performance gain over CS2. If such results come through as strong in other features of CS3, that will be great.

This activity also proved useful in assessing my overclocked system a bit further, as I monitored core temperatures using the Intel Thermal Analysis Tool. The temperatures of the two cores peaked to about 75°C during the Radial Blur, with processor loading indicated at about 99%. Running a 100% load test with the Intel TAT ran core temps up to 82°C, triggering some thermal throttling. That’s hotter than I liked seeing, even though the CPU temp averaged about 20°C cooler. So, I dropped down to 3.5GHz, tightened the memory timings a bit more to 3-3-3-9-4, and reduced the CPU Vcore voltage from 1.42 to 1.39v. Rerunning the 100% load tests, the temps were down to 75°C, for a nice improvement if still on the hot side. During Radial Blur benchmarking, I think core temps were not peaking over 65°C. I suspect the big reason for temperature improvement was the lower Vcore, so I may explore that further once I reread what the operating specs are for the Core 2 Extreme.

Daryl
LH
Lawrence_Hudetz
May 1, 2007
Daryl, you may be interested in this site:

<http://developer.intel.com/design/index2.htm>
DP
Daryl_Pritchard
May 1, 2007
Hi Larry,

Thanks for the link. Funny thing is, that while I didn’t go directly to that page, I’ve been to others today that one might get to from there. I’ve been curious to read more on the E6800 specs, mostly in terms of the operating limits.

Daryl
LH
Lawrence_Hudetz
May 1, 2007
Do you know if the temp tool is reading core or case?
DP
Daryl_Pritchard
May 1, 2007
I’m using Speedfan to monitor both core and case temps. The Intel TAT monitors core only though, for what I can tell. After being idle all night and my just sitting down to check e-mail, I’ve got a case temp of 39°C and core temps of 44/47°C. Hmmm…case just dropped to 31°C as I began typing, probably the fan kicking in as the system woke up fully. I ran TAT for 2.5 minutes at 100% workload on both cores before the thermal monitor went active on one core at a temp of 81°C per the TAT monitor; case temp was 61°C at the time and the other core was 79°C. Speedfan was reporting temps 1-2°C higher and with the test stopped, recovery to case 33, cores 45/48 was pretty quick. It was interesting to see Vcore dropped from 1.38 to 1.34v as the workload peaked…not sure if that’s anything significant or not. Guess I should get out of here to work.

Daryl
RT
Randy_Triplett
May 1, 2007
Daryl,

Wow, Did you see that 40% improvement with CS3 over CS2 in Vista as well as in XP? And with 2gb ram?

I’ve been hearing a lot about how fast CS3 is but I was hesitant to believe the claims. Do you think there is as much real world improvement?

GM, I’m still pricing parts but I will for sure post them back here when I get them. I’m building this to show my nephew how to build a pc, (that’s my excuse to my wife,) so it may be several weeks before I get the thing built.

Randy
GM
gordon_morris
May 1, 2007
Randy,

I’ll look forward to seeing your component list. I’m surprised that there is little information on the best PC for running Photoshop/CS2/CS3. I’d think that Adobe could pull it together easily (they’ve been posting a minimum requirement on their product for years), but they seem strangely silent. Must be some marketing thing I don’t grok. I finally found this info on the Dell site, but their recommendation is not tuned for Photoshop per se. (boy its hard to find)
< http://www.dell.com/content/topics/global.aspx/solutions/en/ wsadvisor?c=us&cs=555&l=en&s=biz>

GM
JJ
John Joslin
May 1, 2007
I’m surprised that there is little information on the best PC for running Photoshop/CS2/CS3.

Ideal specs can vary considerably, depending on the type of work being done in Photoshop.
LH
Lawrence_Hudetz
May 1, 2007
Run everything, therefore run anything.
DP
Daryl_Pritchard
May 1, 2007
<delete>
DP
Daryl_Pritchard
May 1, 2007
Randy,

Here are the results of my radial blur benchmark timing, excluding the tests with Ready Boost which were similar to those without. The times shown here are the averages of 5 test runs each. Is this real world? Oftentimes, I think performance gains are so easily lost to other things we might waste time doing, that the point is moot. However, a 40% gain is nothing to ignore and if more routine tasks in PSCS3 demonstrate noticeable performance gains, then the cumulative effect over a long period of use may well be significant. Beyond that, just the perception of time saved is often a panacea to us when we feel we’re in a rush to get something accomplished. 🙂

Daryl

<fixed>
8 bits/ch 16 bits/ch
CS2 CS3 %Gain CS2 CS3 %Gain
WinXP 40.3 25.4 37.0 44.1 31.0 29.7
WinVista 42.4 26.8 36.8 46.5 32.0 31.2
</fixed>

System Specs:

ASUS P5W DH Deluxe Motherboard
Intel Core 2 Extreme CPU overclocked @ 3.6GHz
2GB Corsair TWINX2048 DDR2-1066 Memory (CAS 3-3-3-10-4)
Drive #1 (System): 150GB Western Digital Raptor SATA150
Drive #2 (WinTemp): 150GB WD Raptor
Drive #3 (PS Scratch): 750GB Seagate Barracude 7200.10 SATA3 PS Memory Allocation: 65% (Approx 1GB)
Vista Ready Boost Cache: Sandisk 2GB Extreme III CF Card

Radial Blur Benchmark Test Action:
1) Create New 8×10-inch, 300ppi, 8-bit image
2) Add Noise: 400% uniform monochromatic
3) Radial Blur: Amount 100, Spin, Best quality
4) Close File without saving

Start Time: Upon click of action "play"
Stop Time: Upon seeing image closed
LH
Lawrence_Hudetz
May 1, 2007
% gain or time?
DP
Daryl_Pritchard
May 1, 2007
% gain…but damn…is there a way to format this neatly?
RT
Randy_Triplett
May 1, 2007
Hey Daryl,

I’m really leaning hard toward that Asus board you’re using. Whenever I see 435 mostly good reviews on Newegg vs 35 overall for the Intel, well, that shows that the Asus has a good following. That and the fact I’ve had really good luck with Asus boards.

Of course I’ve never ever had an Intel board…

Anyone want to discourage me from getting the ASUS P5W DH DELUXE ?

Only bad thing I can say about any Asus I’ve had is one onboard fan went loud on me once. Then again I cut that fan loose and the machine is still running, and is quieter too I might add. Did I mention I like quiet? 🙂

Randy
DP
Daryl_Pritchard
May 2, 2007
Randy,

I think that mobo is a good choice if the features meet your needs. I really liked the feature set and layout although I don’t use the Digital Home remote or other DH features (dynamic fan/voltage monitoring for quiet operation). I’ve also found that updates for Windows Vista are currently only an essential few. I’ve yet to see updates for the ASUS PC Probe II, DH Remote, Ai Boost, etc., but I suspect you can do without them pretty easily given your objectives for a quiet, non-overclocked system. Speedfan is a good substitute for PC Probe II. The main thing I suggest if you go with this board is removed the cheesy sheet aluminum "cap" over what I think is the northbridge. All it does it boast the ASUS name while hindering cooling and it is attached with a small amount of glue, so it is pretty easily removed with a pair of pliers, pulling carefully on one edge to separate it from the heatsink and progressing from there.
DM
dave_milbut
May 2, 2007
is there a way to format this neatly?

yes. type it in notepad and make it look right. post it here in between

<pre> and </pre> tags. that will preserve formating/spacing.
DP
Daryl_Pritchard
May 2, 2007
Dave,

Thanks…I did in fact edit the text in Notepad yet didn’t know of the "pre" tag. I tried "mono", "fixed", "font face=<some monospace font>", etc. I think it was "fixed" that used to work before the forums were moved a few years ago.

Daryl
LH
Lawrence_Hudetz
May 2, 2007
The northbridge needs a heat sink, as does the south bridge, though not nearly so extensive.

The northbridge should have at least fins.

But then, they may be using other than Intel chips.

Please proceed carefully with messing with these devices. Excessive stress on the chips can fracture them.
DP
Daryl_Pritchard
May 2, 2007
By your comments Lawrence, I’m sure it was the Northbridge which ASUS stupidly put that thin plastic-laminated, aluminum cap over. Visiting the Anandtech thread < http://forums.anandtech.com/messageview.aspx?catid=28&th readid=1909360&arctab=y&frmKeyword=&STARTPAGE=8& amp;FTVAR_FORUMVIEWTMP=Linear> where I discussed the board setup with others who were using it, I reported that removing the cap was good for a 5-10°C improvement in the board temp.
LH
Lawrence_Hudetz
May 2, 2007
That’s weird! But be careful about it as the chip can be fractured. Then the board is toast.

Which brings up another point about cooling. I just installed an nVidia graphics card with passive cooling, and the principal heatsink is under the board as it sits in the computer.

Huh? Convection cooling and the heat goes into the sink as it tries to escape?

Makes me doubt I know anything about anything!
DM
dave_milbut
May 2, 2007
Makes me doubt I know anything about anything!

I get that feeling all the time. Don’t worry, you learn to live with it. 🙂
LH
Lawrence_Hudetz
May 2, 2007
At age 70? 🙂
BL
Bob Levine
May 2, 2007
I’ll let you know in 20 years. 😉

Bob
DP
Daryl_Pritchard
May 2, 2007
Lawrence…ditto, Ditto, DITTO on the strange idea of putting fans or heatsinks on the underside of a hot graphics card. You’d think the cards would be designed for the components to be on the upper surface.
LH
Lawrence_Hudetz
May 2, 2007
20 years, ‘eh? Goes by quickly.
I well remember celebrating my 50th birthday. i think I can still smell the candles (and the dope!).

Maybe we should redesign the cases to mount the boards horizontal. The test platforms are horizontal and imo, they run better. Besides, who ever heard of a vertical platform?

Actually, seems to me that horizontal cases are around.

Off to the races…..
ND
Nick_Decker
May 2, 2007
Randy, in your initial post you mentioned that you had a "7600GT fanless" card. Can you be more specific about the brand and model number? I may be upgrading my video card, and like you, I don’t like no stinkin’ fans running that aren’t necessary.

Thanks.
RT
Randy_Triplett
May 2, 2007
Nick,

It was a GIGABYTE GV-NX76T256D-RH GeForce 7600GT 256MB GDDR3 PCI Express x16 Silent Pipe II. That’s a long description pulled straight from the Newegg invoice.

I bought it about a year ago and it’s been bombproof with DirectX 9 games and also works great with Ubuntu. (as I’ve already mentioned, I don’t tweak settings.)

And Lawrence, I too have wondered about the heatsink on this card, as it is also on the underside of the card. Go figure…

Randy
MD
Michael_D_Sullivan
May 2, 2007
There are a couple of the 7600GT fanless cards. Mine is by Asus, "EN7600GT SILENT". A review from Oct. ’06 is available here: <http://www.viperlair.com/reviews/video/asus/7600gt/>. I’m very pleased with mine.
RT
Randy_Triplett
May 2, 2007
BTW All,

When looking back at the order that included that 7600GT, I saw the Antec case I used and will use again on this build.

< http://www.newegg.com/Product/Product.asp?Item=N82E168111291 66>

This is a beautifully made case with suspension mounted drives and modular cable routing. I was really impressed with it and the very quiet power supply and the very quiet fans.

Worth a look.

Randy
DP
Daryl_Pritchard
May 2, 2007
That looks like a nice case Randy, but it’s missing the fancy-schmancy "custom front panel light beam projector" of my Gigabyte Poseidon < http://www.newegg.com/Product/Product.asp?Item=N82E168112330 06> case….as if I ever even notice it on my carpeted floor! Ha!

I’m very happy with the Gigabyte case and like the styling of it, but it probably is a little noisier since it has a ventable side panel (vent interchangeable with acrylic window). A plus for me was 5 external drive slots over the more common 4 or even 3 on some mid-tower cases.

With regard to the "modular cable routing" of the Antec case, what are you referring to? It looks pretty conventional to me save for the modular power supply cables, and maybe that’s what you mean. If so, I agree…that’s one thing I do miss, as I had to just securely tie up the unused power cables of my Seasonic p/s.

Daryl
ND
Nick_Decker
May 2, 2007
Thanks, Randy and Michael, for the info on the 7600GT cards. Unfortunately, I’m still using a motherboard (Asus) that needs AGP. Oh well, maybe next time.
RT
Randy_Triplett
May 2, 2007
What is this "AGP" of which you speak?
MD
Michael_D_Sullivan
May 2, 2007
Here’s an AGP version of the 7600GS:
< http://www.newegg.com/Product/Product.aspx?Item=N82E16814121 064>

There is also, apparently an AGP version of the 7600GT:
< http://www.firingsquad.com/news/newsarticle.asp?searchid=138 07>

AGP is the graphics bus that was most popular before the introduction of PCI Express.
ND
Nick_Decker
May 2, 2007
Hey Michael, thanks for the links!
DP
Daryl_Pritchard
May 5, 2007
I recently found that DriverHeaven.com has a Photoshop benchmark that exercises various filters and other commands in an effort to establish what they call a "real world" benchmark. True or not, at least it is something to use. So, using both it and the Radial Blur benchmark I’ve mentioned before, I set about benchmarking PS CS2 vs. CS3 under Windows Vista, in what I call an "RBP" (Reduced Background Processes) state and a normal boot state with all usual background utilities and such active as I’d normally have them. The results with the DriverHeaven benchmark show CS3 offering around a 9.5% gain over CS2…notably less than the >30% gain demonstrated by the Radial Blur benchmark alone.

The results are at: <http://ambress.com/photoshop/psbench_vista.htm>

Regards,

Daryl
H
Ho
May 5, 2007
Daryl, give the Retouch Artists speedtest a whirl. I like it better than DH:

<http://www.retouchartists.com/pages/speedtest.html>

You can also compare your results to those who use than *other* OS. 🙂
I
ID._Awe
May 5, 2007
Well I got my case here: Wang Tsann <http://www.wangtsann.com/index_files/Page6580.htm>

Got the cool ‘C’ case. This case has two separate sides for the MB & drives with a fan behind the drive cage.

I use quality fans (4) that suck air in from the front and blow air out the back (just switch the fans around). Heat has never been a problem.

I have also been using heavy duty power supplies (550 watts rated for use in hospitals), just so when you are really crunching serious numbers there is a large power overhead.

5 years ago when I bought it, everyone thought I was missing some screws, but it has brought 5 years of headache free operation.

ABITVP6 with dual 1ghz CPUs, 2GB RAM, & 5 hard drives all running at a matched 133Mhz. Still running strong, although I am now lusting for a hardware update.
DP
Daryl_Pritchard
May 5, 2007
Ho, thanks….I’ll look into that test when I get the chance.

Daryl
DP
Daryl_Pritchard
May 5, 2007
Ho,

I’ve not updated my results page with the RTA benchmark, but the results I’ve measured seem rather awry, slower than the top 5 results I see posted for a Windows PC, albeit only a second off the 4th & 5th-place spots. And CS3? Ewwww! This is the first benchmark that makes CS3 look poor in performance compared to CS2.

RTA Benchmark CS2: 60.41 secs, CS3: 85.62 secs

Out of curiosity, given my system (Core 2 Extreme @3.5GHz) didn’t outperform the #1 spot, a Core 2 Duo overclocked at 2.72GHz, I’m going to reboot to spec and rerun the benchmark. This is not a matter of bragging I’ve got a fast system, just that something seems very wrong for it not to give better results in contrast to those top 5. I’ll also check it out under WinXP…maybe Vista is the culprit here, but that hasn’t borne true for the other benhmarks.

Daryl
LH
Lawrence_Hudetz
May 5, 2007
Wouldn’t surprise me!
DP
Daryl_Pritchard
May 5, 2007
Hi Lawrence,

Well, I’m wondering now if it is my overclocking…or perhaps CS3 gremlins. I’ve yet to reboot but decided to relaunch CS3 while running Speedfan to monitor temps during the execution of the RTA Benchmark. That 85.62 seconds dropped inexplicably to 47.35 seconds! And the next run?….29.56 seconds! Over a series of 15 runs, observing that Speedfan active or not had little effect, I timed the benchmark at a best of 28.35 seconds to a "this series" worst of 49.41. Only five timings were 29.5 or faster. I also noticed on those slower runs that PS CS3 reported several instances of "Not Responding", always just prior to a point in the test where a gaussian blur is performed.

Moreover, and maybe this has been reported with CS3…I noticed that upon closing the test image without a save, there were instances where it closed immediately, but others where it took anywhere from 5 to 18 seconds to close. What’s up with that?

I’ve heard of memory leaks but am not familiar with what that is about, how to confirm it, etc. but wonder if that is what is occurring. I see nothing the looks awry if I monitor the RAM utilization (Vista has great monitoring tools in contrast to XP).

Time for a spec run…

Daryl
LH
Lawrence_Hudetz
May 5, 2007
Keep the clock speed in check and try again.

I would like to know your results.
H
Ho
May 6, 2007
Daryl,

It’s been quite a while since I ran the benchmark, but as I recall it creates a large enough file to hit the Scratch disk so it’s not just a processor test. How many History states do you have set?
DP
Daryl_Pritchard
May 6, 2007
Running 1 history state Ho. One thing odd…not sure if a CS3 issue, Vista, or the benchmark having problems, but in a few cases I saw my PS scratch file hit 9GB in size, when normally it didn’t exceed 3.5GB. The slow timings weren’t all connected to that scratch disk peak, but when it did occur, the timings were probably worst…makes sense. I’ll post data when I get a chance to check out performance on WinXP, where even running the benchmark without paring down my processes already looks better with both CS2 & CS3.

Daryl
DP
Daryl_Pritchard
May 6, 2007
Still messing with things…still seeing odd behaviors that I can’t seem to attribute to my overclocking, but I did come across one item that surprised me:

My system normally runs with 4GB of memory, but currently is at 2GB while I await replacement modules from Corsair. Well, I overlooked (or thought it unnecessary, to be more correct) booting WinXP without the /3GB switch, and was surprised to find PS CS2 reporting 2661MB available when only 2GB was installed. I had thought that perhaps even with the /3GB switch active, the actual installed memory amount would be detected and reported by the O/S to whoever needed to make use of that info. Apparently not. It seems more as if the /3GB switch just says "always report 1GB more than actually detected".

Part of what led me to catching this, was an observation over 3 consecutive runs of the RTA Benchmark on PS CS2, that the PS scratch file peaked at 7.5GB in size, and timings were atrocious…2-4 minutes per run! I’d seen an occasional 9GB peak file size under CS3, but that was while running Win Vista and without any boot option mods since I’ve yet to learn what they are and didn’t need them yet.

I’ve run my system now at 3.5, 3.24, and a BIOS default speed of 2.13 GHz (I thought the default would be the Core 2 Extreme spec of 2.93GHz, but nope), and some of the odd timing results I’ve seen under Win Vista and most notably with PS CS3 seem to persist. So, that leads me more to suspect software issues in CS3 and/or Win Vista than any overclocking issues. What really puzzles me though is, on the same image that is closed and reopened for each run of an action, why would execution times of that action vary rather widely unless something is amiss in Photoshop? No doubt swings in the size of the scratch file would certainly play a role, and yet I don’t understand why they would fluctuate for the same image and action either. There does seem to be a consistency of the first run time being the longest for the benchmark action, and that is even when a Purge All or file close and reopen is performed…almost as if that first run "preconditions" PS to run better. Odd.

Daryl
LH
Lawrence_Hudetz
May 7, 2007
What’s the FSB speed of your system?
LH
Lawrence_Hudetz
May 7, 2007
If it’s 1066Mhz, 2.13 GHz requires a multiplier of 8 whereas at 2.93 GHz, it’s 11! I don’t know of any BIOS that goes that high. If it’s 1333, the multiplier is 9 for 2.93. Actually, 2.999 or 3Ghz, FAPP. A 6x multiplier from a 1333 Buss speed is 2Ghz. 6x setting is normal.
DP
Daryl_Pritchard
May 7, 2007
At which point in time Larry? ha! I’ve been at several different FSBs today. I’ve explored all of 266, 300, 330, 350, and 360MHz core speeds, the 266 being the only exception of not running a 1:1 DDR:FSB ratio, as that was the BIOS default setup. As I’ve mentioned before, this ASUS motheboard with the Intel 975 chipset just doesn’t seem to like anything much over 360-370MHz for the core FSB when running a 1:1 DDR:FSB ratio as I’m aiming for. While I was running OK at 3.6GHz today I decided to leave the system as I’ve had it recently, at 3.5GHz, dropping Vcore just a little also, all in the interest of running a little cooler…not much, but anything helps. So, as things are right now, I’m running a 350MHz core FSB for an effective quad-pumped FSB of 1400MHz. Memory is clocked at DDR2-700MHz and 3-3-3-9-4 CAS timings.

Interestingly enough, it seems those odd peaks I observed in the PS scratch file size while benchmarking today were the result of that misleading /3GB switch while having only 2GB in my system. Once I booted without that option, I don’t believe I ever saw the awry scratch file sizes again.

What is peculiar and yet also reasonably consistent, is that whether using WinXP or WinVista, CS2 or CS3, that Retouch Artists Pro benchmark always gives a slow timing measure on its first run. Close the test image and reopen it, and eventually the results show a marked improvement. Oddly, CS2 will begin responding better on the benchmark after the 2nd or 3rd run, while CS3 usually requires 4-5 runs. Given that the image file was closed and reopened each time I ran the benchmark, I’d expect all working areas to be flushed and not giving PS the benefit of some old, unchanged data to work with. So, I just haven’t a clue why I saw what I did.

Another thing puzzling about the RTA Benchamrk is they specify setting the RAM allocation to 100%, which I’ve never seen recommended yet sure enough, it did give better results than a more typical 60-70% allocation. I’ve yet to update my web page with the results, but here they are with the timings from multiple runs:

RETOUCH ARTISTS PRO BENCHMARK RESULTS:

Core 2 Extreme CPU @3.6GHz
FSB 1400MHz
Memory @ DDR2-700 for 1:1 DDR:FSB ratio
CAS Timing: 3-3-3-9-4

PS Cache Levels: 4
History States: 1
Memory Allocation: 100%

Boot State: All non-Microsoft services disabled.
All Startup items disabled except video and mouse drivers All Start Menu items disabled.

Windows XP Results
==================
3.6 GHz CS2: 55.41 32.75 29.50 30.54 30.54 30.57
3.6 GHz CS3: 48.83 48.86 48.27 29.16 28.28 27.79 28.04 27.99 28.05

Windows VISTA Results
=====================
3.6 GHz CS2: 59.74 37.23 34.96 32.00 31.35 31.46
3.6 GHz CS3: 49.10 49.18 48.98 47.58 28.60 49.41 30.86 31.73 30.07

Regards,

Daryl
LH
Lawrence_Hudetz
May 7, 2007
If your clock speed is 350 your multiplier is 10, which, I believe, ia all the BIOS usually allows for. Maybe extremem boards and cpu combos will go farther.

FSB is, like you said, 1400.

Pretty good!
DP
Daryl_Pritchard
May 7, 2007
With some earlier version of my board’s BIOS, which I just updated again today, I had seen multipliers up to 13 available. I don’t know if that was for all clock rates or not, but certainly the low ones, such as 266MHz. Now, it appears that 11 is the limit. That would be good for 3.85MHz at the moment, but I’d need better cooling than my air cooler I think. I did try booting to 3.85MHz after boosting Vcore to 1.425v, but it would never get past the initial boot stage for Windows. No surprise there…I’ve had 3.75GHz before, but it was unstable.
LH
Lawrence_Hudetz
May 7, 2007
You are pushing hard there, Daryl!
H
Ho
May 7, 2007
Do you close Photoshop between runs? If not there is some likelihood that some results are not being cleared from cache (scratch). Some people advocate reboots between benchmarking runs.
LH
Lawrence_Hudetz
May 7, 2007
My friend with the core 2 is experiencing general slowdowns as time goes on. I was there last night showing a function and I was surprised at how long it now takes to come up and run a filter.

Speed is ephemeral!
DP
Daryl_Pritchard
May 7, 2007
Ho,

I had wondered about that. I saw where the setup instructions for the RTA Benchmark specified doing a reboot after you set up PS to the settings they use, and I did that. But I didn’t do it for the repeat runs. For those I only closed the test image and reopened it, but did so while monitoring the scratch disk file (the PS temp file), which would always be set to the same size each time the image was opened.

Perhaps my data isn’t valid, but for me to obtain the initial timings that I did for both CS2 and CS3, under both XP and Vista, seems very awry if compared to the top results posted of 37 seconds on an overclocked 2.72 GHz Core 2 Duo for CS2. That is, if my 3.5GHz system can only do the benchmark on CS2 under WinXP in 55 seconds, an 18-second deficit compared to this notably slower-clocked system, then something just seems very wrong to me. On the other hand, if the data submitted for that system was also a "best of" time for a series of successive runs as I did, then the numbers look more as they should, with mine being about 6 seconds faster. But, whether 1 second or 6, that doesn’t matter….theoretically at least, my system should be faster. If not, then I’d like to figure out why, which isn’t likely.

Daryl
RT
Randy_Triplett
May 7, 2007
Daryl,

Step away from the Benchmarks…

Really, Dude, you have a very fast computer.

Maybe it’s time to be happy with it… 😉

Randy
DP
Daryl_Pritchard
May 7, 2007
Randy,

Yeah, I hear you…I should definitely be doing more productive things than running the same benchmark under umpteen different configurations! Ha! I’ve got a feeling that tweaking becomes a bit addictive, but it has also been somewhat enlightening to me, as I’ve started paying closer attention to internal temperatures. For example, while I was running at 3.6GHz with a Vcore (CPU voltage) of 1.425v with no problem, I did find that Intel’s Thermal Analysis Tool would send the core temperatures up to 82°C, triggering the thermal monitoring protection of the Core 2 Extreme, if I loaded both cores at 100%. The settings I was using were largely derived from what another person found success with in their overclocking on the same CPU/mobo combo. But, I found at running at 3.5GHz with a Vcore of 1.34v resulted in a noticeable reduction of those core temps at peak loading, yet obviously 3.5 vs. 3.6GHz is no significant performance loss. I’d still like to see if I could cool things down a bit more, but I’ll most likely pursue that via a swap of my current cooler for a more effective model I’ve just learned of, the Enzotech Ultra-X. No rush there though.

Daryl
RT
Randy_Triplett
May 7, 2007
Daryl,

Yep, I know, man I’ve been there too. You’ll tweak and twist and poke at the thing and just when you finally are convinced you have it exactly the way you want it, CS4 will come out and require eight cores.

But it is fun and informative to mess around with all that stuff. I’m just giving you a hard time.

Randy
RT
Randy_Triplett
Jun 1, 2007
Well, Boys and Girls, I’m typing this on my new system. It’s not quite finished but here’s what I have so far:

Intel Core 2 Duo E6600 Conroe 2.4GHz LGA 775 Processor Model BX80557E6600 – Retail $226.00

EVGA 122-CK-NF66-T1 LGA 775 NVIDIA nForce 650i Ultra ATX Intel Motherboard – Retail $109.99

WINTEC AMPX 2GB (2 x 1GB) 240-Pin DDR2 SDRAM DDR2 800 (PC2 6400) Dual Channel Kit Desktop Memory Model 3AXT6400C5-2048K – Retail $88.99

Sony NEC Optiarc 18X DVD±R DVD Burner With 12X DVD-RAM Write Silver E-IDE / ATAPI Model 7170A-0S – OEM $32.99

Antec Performance One P150 White Steel ATX Mini Tower Computer Case 430-Watt ATX12V v2.2 Power Supply – Retail $149.99

WDC WD2500JBRTL 250gb SATA 300 $59

nVidia GeForce 6600GT PCI-E (re-purposed)

"Budget" RAM and mobo were highly recommended on several forums and so far so good. (Mobo doesn’t do SLI, but neither do I.) Video card was just lying around and I thought I’d try it rather than rob my gaming pc. So far it looks great on my dual Syncmaster 204b’s.

I haven’t added my second drive yet, which will be a 500gb SATA 7200 rpm drive. Despite many of your recommendations, I am still only an enthusiast and economy is more important to me than speedy Raptors. Also, I will still place my scratch disk on the first small (10gb,) partition of the secondary drive. I’m keeping all of your advice in mind, but this solution has always worked well for me.

I’ll get my secondary drive at the first of next week and I am anxious to test out the new system!

Please recommend some benchmarking tools I can play around with. Also, I’m looking for a new system "analysis" tool like my beloved Everest (Lavalys) that is not being updated any more.

Thanks again for all of your advice on this subject.

Oh, and the build went smoothly and the nephew is ready to build his own now. Pass the torch!

Randy
DP
Daryl_Pritchard
Jun 2, 2007
Hi Randy,

Congrats on the new system and getting a nice-value setup.

Hmm…Everest isn’t updated anymore? I just recently downloaded an updated version for Vista, but perhaps you’re referring to the "lite" free version? I liked Everest well enough that I bought Everest Ultimate some time ago. I’ve benchmarked my system using CrystalMark, Everest, and SiSoft SANDRA to name a few, and SuperPI for a rather trivial benchmarking. All this was done rather recently also, as I made another system change, installing a Thermalright Ultra-120 Extreme heatsink and some nice quiet, but high-flow SilenX fans. This came as the result of our earlier discussions here, and my observations of system temps under max loading of both cores with the Intel Thermal Analysis Tool. With the mods I’ve made, which also included removing all heatsinks on my ASUS mobo, cleaning the chip surfaces, and prepping them with Artic Silver 5, I have now dropped my system temps under max loading by about 6-8°C and during more routine use by about 10-12°C. Plus I reduced the noise levels since the new fans are quieter than my Zalman cooler’s fan was. I’ve updated my PC info on my website if you want to see the changes and the benchmarks, at <http://ambress.com/pc>.

Note that while I did quieten and cool things down, I didn’t really get any more performance out of the CPU. Previously, the max load testing would send my CPU cores to the thermal shutdown levels at 3.6Ghz, so I pulled back to 3.5GHz and dropped my Vcore some, and that prevented the shutdown at max loading. Now, I’m back to 3.6GHz while having an even wider safety margin on the temperatures, so I’m happy to stay put. The only thing I may yet change is to swap to a passive heat sink for my graphics card and get rid of the noisier OEM fan. No rush there though, as this is just getting into overkill mode where the noise at present doesn’t bother me at all.

Glad to hear all went smoothly with the build…fun stuff, ain’t it? 🙂 Best way to learn, that’s for sure.

Daryl
RT
Randy_Triplett
Jun 2, 2007
Daryl,

You’re a maniac. Replacing all the thermal paste on the board? I guess it wasn’t that much trouble, but frankly I never would have even thought of it! Amazing.

That’s a helluva radiator on your cpu now too. How did you like the fit?

I must admit, I have been using the free version of Everest for so long, I didn’t even realize there was a paid version. I’ll have to try it.

Randy
DP
Daryl_Pritchard
Jun 2, 2007
Randy,

Yes, this is the first PC I’ve gone to such lengths with and it is a bit obsesssive but obviously there was at least some payoff for the efforts. Replacing the thermal paste was quite easy and was prompted by posts I’ve read from others doing the same with good results. When I saw how only the edges of the southbridge had paste on them, I was glad to better prep it with the Arctic Silver. Maybe it doesn’t really need it…not sure how hot it gets in contrast to the small northbridge chip which has the more substantial heatsink assembly. I think I’m done now…no plans to lap the CPU or CPU cooler surfaces as some do, to ensure perfectly flat and smooth contact.

I double-checked my case dimensions before buying the Thermalright cooler. It was 165mm tall to my case’s 200mm width. There was enough room to spare even with the motherboard elevation factored in, but not much. I have a 120mm fan fastened to the screen window on the chassis side panel and thought it might have to come off or be swapped for a smaller fan, but the cooler just clears the top edge by about 5mm and the tips of the heat pipes end up about 14mm from touching the side panel. I couldn’t ask for it to be much closer than that! I thought I might quieten the case more by replacing the screen panel with an acrylic window supplied with the case, but I can’t find it. I wasn’t sure if it doing that could actually improve cooling by forcing all air to come in from the front of the case, but simply covering the side panel temporarily and disconnecting the fan did show a slight rise in the system temperature without any real benefit to the hard drives, so I’ll leave things as I have them currently.

For only $30 US, Everest Ultimate seems a nice utility to have for anyone interested in probing their system’s details.

Daryl
LH
Lawrence_Hudetz
Jun 3, 2007
Daryl, it’s incorrect to say you have widened the safety margins on the temps. You haven’t. What you did was to increase the thermal efficiency so that at a given power output, the temperature is lower. If your temperature reaches the same value at which the cpu failed, it will fail again.

Widening the temperature margins means to me that instead of failing at say 70C, you can now operate up to 80C. So, you actually increased the dissipation margins, or decreasing the thermal resistance.
DM
dave_milbut
Jun 3, 2007
engineers! sheesh! 🙂
DP
Daryl_Pritchard
Jun 3, 2007
Lawrence,

I agree with you but just looked at this a bit differently. I think of the safety margin as the delta between the Tjunction and the peak core temp of my system at maximum loading. My prior configuration was less thermally efficient, so the "safety margin" (and perhaps I should’ve used a different phrase) was narrow. I’m not sure where thermal shutdown occurs for the X6800, which has an 85C Tjunction, but let’s just say it is 82C and my old configuration would peak at 80C. In my way of thinking, that was a 2C "safety margin" which I’ve now widened to 6C, since the hottest core is peaking at 76C.

The best way of putting it though would simply have been to say that "I improved my PC’s thermal efficiency such that it now runs cooler under maximum CPU loading than it did before, with only a minor performance gain." 🙂

Or, in shorthand…. "My PC is better than it was."

lol @ dave!

Daryl
LH
Lawrence_Hudetz
Jun 3, 2007
I don’t disagree with the margins obtained, only the presentation.

The best way of putting it though would simply have been to say that "I improved my PC’s thermal efficiency such that it now runs cooler under maximum CPU loading than it did before, with only a minor performance gain."

Exactly!:-)

C’mon dave. Some software presentation that doesn’t sit well will get you off your rocker too! 😀
DM
dave_milbut
Jun 3, 2007
😉
RT
Randy_Triplett
Jun 3, 2007
You guys need to get a room…with a chalkboard…and work this thing out… 😉
LH
Lawrence_Hudetz
Jun 3, 2007
I only get a room with a woman.
DP
Daryl_Pritchard
Jun 3, 2007
And that is the single smartest comment you’ve made yet Larry! lol

As for chalkboards, phooey on ’em…it’s when theory is presented that I lose interest. Hands-on is much more fun. When I see a puff of smoke, then I know "well, there goes that theory!" ha! Luckily, no smoke yet.
LH
Lawrence_Hudetz
Jun 3, 2007
And that first comment is the smartest I’ve seen yet Larry!

Geeze, do I come across that dumb?:D

Luckily, no smoke yet.

Uh oh!

Master Retouching Hair

Learn how to rescue details, remove flyaways, add volume, and enhance the definition of hair in any photo. We break down every tool and technique in Photoshop to get picture-perfect hair, every time.

Related Discussion Topics

Nice and short text about related topics in discussion sections