Mates,
First, thanks to PassMark for an intelligent, intuitive application. Being able to filter among tens of thousands of benchmark tests and compare has helped clarify the relationship of CPU, video card, RAM, and disk parameters to the overall system performance.
And, in a flurry of use of my trial, I made significant performance improvements to two of my three old bangers >
1. Dell Precision T5400 (2010)> 1X Xeon 5460 (3.16Ghz, 4 Cores), 12GB DDR2 , Western Digital RE4 500GB, Nvidia geforce 285 GTX (1 GB), M-Audio 2496 soundcard. Win 7 Ultimate 64 bit. AutoCad 2011, ArchiCad 14, Sketchup, Adobe CS3, Corel Technical X5, WP X4
2. Dell Optiplex 740 (2006)> 1 X AMD Athlon 64 X2 4450B (2.3Ghz, 2 cores) 6GB DDR2, Seagate Barracuda 500GB, Nvidia Quadro FX580 (512MB) Win 7 Ultimate 64 bit, AutoCad 2011, Revit, Inventor 2011, Solidworks, CS4, Maya 2011.
The first computer I tried for benchmark testing however was one of my semi-retirees >
Dell Dimension 8400 > 1X Pentium 4 630 (3.00Ghz, 64-bit, 1 core), 3GB DDR2 RAM, Western Digital 80GB, Nvidia Quadro 570 (256MB) XP Home and Business 32-bit. AutoCad 2007, Corel Graphics Suite 12, WP X4. This is 2003 and I think among the first 64-bit capable Dells (Prescott). I believe when I ordered it, there were not quite yet any 64-bit OS or applications.
> which scored an overall 452, which I discovered was quite good for a single core Pentium and was quite good for a Dimension 8400. The 2D score of 444 was encouraging and going through the results for the Quadro 570 in various systems, discovered that card appearing in a surprising number of good performing workstations.
I then tried the 8400 with a different video card, a Nvidia geforce 9400 GT, 1GB GRR2. This time the score was 427 as 2d dropped from 444 to 225 in favor of 3D going from 261 to 329. That I think reveals the difference between the graphics oriented Quadro and the game-looking geForce.
In the first run of the Dell Precision T5400, the overall score was a satisfying 1852- and read as so much more impressive than the Dimension's 452. The disk score showed the WD RE4 Enterprise drive to be working well- those have 64MB cache. However, the 2D score of the massive, hot, 200W 384-bit, 1GB, 240 CUDA core GTX 285 was only 300- not as good as the Dimension's 444 from a 256MB, 16 CUDA card! The 3D of the 8400 was 261, for the Precision, it was a respectable 2208.
> It seemed impossible that the 2D of the T5400 was so poor relatively. > Something was seriously hobbling the works!
When testing the Optiplex 740, the score was 776 and like the T5400 the 2D results were mysteriously dismal- 380- but again, still better than the 300 for the GTX 285 in the Precision!
It occurred to me then that the problem must be some extreme graphic drain on the system and it took quite a while to find it. I had had an unusual problem recently in that when I installed Corel Technical Designer X5 on both the Precision and Optiplex, PhotoPaint X5 would not open multiple files. It would open the first selected file then present "I/O read error" for each of the other files.
After three re-installs of the suite- 7GB all in, numerous registry hunts and, stupidly, two 7-8 hour repair installs of Win 7, I discovered that it was my use of the Windows Classic theme in Windows 7 Ultimate that caused this error ! By trying to use the least resource-intensive setup, I had caused 20+ hours of needless, frantic effort to try and get a fairly expensive-and frequently used software to work. Corel Graphic Suite 12 was not happy on Windows 7 and running xp mode had graphic penalties- it switched to generic drivers!.
The solution to the multiple file issue, which I read on my 490th search on this subject, was to "switch from Windows Classic to an Aero Theme". I did so, choosing the one with the least number of helpful fuzzy bears.
Wow!
What I didn't realize until seeing these benchmark tests- thanks again- was that the Aero theme's transparent windows, the fading and the animations of tilting windows up and down must be seriously robbing the applications of GPU time. Window transparency seemed to be one of the hungriest components.
I sorted out a Windows 7 "Basic" theme- all solid colors- that still allowed opening multiple files in PhotoPaint X5 and ran the Precision T5400 test again. The results were astounding for this one, 5 minute change. The pleasant-sounding 1852 jumped to 2329 which I found is among the best performing single CPU T5400's and outperforming some multiple CPU and Core i7 machines. I can see too in this that DDR3 is noticeably better than DDR2 and SSD's are the future,..
Tellingly, in the new T5400 result, the 2D score jumped significantly from 300 to 583- that reads as almost doubling, 3D rose 2208 to 2320 and the CPU result changed from 4952 to 5252.
All these improvements from turning off the Aero theme!
Lesson> All the fuss I had with Corel over this issue, and all those many hours could have been saved if I had taken the time to run these tests and understand the results- a significant performance improvement for $0 that would have saved 25 hours of frustration.
Observation > Research > Analysis > Experiment
-whereas my technique of screaming at the monitor,..
The results with the Optiplex 740 were also revealing, especially that the 2.3Ghz Althlon 64 X2 4450B CPU was holding the system back. Where the T5400 Quad core Xeon was making a CPU score of 5252, the Dual core Althlon was 1441. The Windows Experience Index of 5.1 told the same story- it's the processor score lowest giving the 5.1. Again the Quadro 580 is doing well- Graphics and Gaming both 6.4.
Based on these two clear messages concerning the Optiplex CPU, I bought ($55) and installed a used Athlon 64 X2, changing the 2.30Ghz, 4450B to a 3.00Ghz 6000+, which I think also doubles the L2 cache to 2MB. (The Xeon 5460 has 12MB L2). By the way, anyone thinking of this upgrade should watch carefully for the different versions of the 6000+, the 285W PS and tiny desktop case of the Optiplex 740 desktop- and other designs apparently object to the 125W version in place of the 89W. The 89W prefix is "ADA" while the 125W can be "ADX" and possibly others. The 6000+ has been over-clocked to 3.2Ghz- supposedly reliably, and the 125W version does not have a wider over-clocking margin than the 89W. There is another 3Ghz 64 X2, the 5800, but I think that one has 1MB L2 cache instead of 2MB in the 6000+. There is also a "3.1" 6000+ , 89W and 65nm instead of 90.
Having learned the Aero theme lesson and with the 6000+ 3Ghz CPU all thermally pasted into the Optiplex 740 (my first CPU change), I ran the Passmark test again and voila, the 776 became 1113! I was quite amazed at the proportion of increase- 40%, and making it among the best performing AMD Optiplex 740's. The CPU improved 1441 to 1909 (+32%), 2D> 225 to 484 and 3D> 380 to 429. It's difficult to separate the effects of the CPU and Aero theme changes, but it seems this is, similar to the T5400, again telling about the disastrous Aero themes that the 2D graphics scores changed so significantly without any other changes.
I'm curious as to whether there is widespread similar experience with Aero themes so blatantly robbing computer performance as well as preventing Photopaint X5 from working properly. The Photopaint frustration was one the worst in my 19 years of fussing with PC's.
I was never a computer hot-rodder before, but the new imperative to do everything in 3D- even CorelDraw 2D is now Designer 3D, AutoCad became Revit and so on, makes having the best graphics system possible. - And with some careful testing and consideration, it seems really noticeable improvements can be made- and economically- to older computers, thanks to this kind of testing/analysis. I suppose new computers by definition are already running nearer the current SOTA, so the old crocks can show bigger gains,..
Sorry to go on for so very long, but I thought the discoveries made were useful and might save others' time and frustration. If you've read this far- thanks and,
Cheers,
Bambi B
Los Angeles
PS Next week, I'll see what my 1998 Dell XPS P3 750, 512MB RAM can do,..
First, thanks to PassMark for an intelligent, intuitive application. Being able to filter among tens of thousands of benchmark tests and compare has helped clarify the relationship of CPU, video card, RAM, and disk parameters to the overall system performance.
And, in a flurry of use of my trial, I made significant performance improvements to two of my three old bangers >
1. Dell Precision T5400 (2010)> 1X Xeon 5460 (3.16Ghz, 4 Cores), 12GB DDR2 , Western Digital RE4 500GB, Nvidia geforce 285 GTX (1 GB), M-Audio 2496 soundcard. Win 7 Ultimate 64 bit. AutoCad 2011, ArchiCad 14, Sketchup, Adobe CS3, Corel Technical X5, WP X4
2. Dell Optiplex 740 (2006)> 1 X AMD Athlon 64 X2 4450B (2.3Ghz, 2 cores) 6GB DDR2, Seagate Barracuda 500GB, Nvidia Quadro FX580 (512MB) Win 7 Ultimate 64 bit, AutoCad 2011, Revit, Inventor 2011, Solidworks, CS4, Maya 2011.
The first computer I tried for benchmark testing however was one of my semi-retirees >
Dell Dimension 8400 > 1X Pentium 4 630 (3.00Ghz, 64-bit, 1 core), 3GB DDR2 RAM, Western Digital 80GB, Nvidia Quadro 570 (256MB) XP Home and Business 32-bit. AutoCad 2007, Corel Graphics Suite 12, WP X4. This is 2003 and I think among the first 64-bit capable Dells (Prescott). I believe when I ordered it, there were not quite yet any 64-bit OS or applications.
> which scored an overall 452, which I discovered was quite good for a single core Pentium and was quite good for a Dimension 8400. The 2D score of 444 was encouraging and going through the results for the Quadro 570 in various systems, discovered that card appearing in a surprising number of good performing workstations.
I then tried the 8400 with a different video card, a Nvidia geforce 9400 GT, 1GB GRR2. This time the score was 427 as 2d dropped from 444 to 225 in favor of 3D going from 261 to 329. That I think reveals the difference between the graphics oriented Quadro and the game-looking geForce.
In the first run of the Dell Precision T5400, the overall score was a satisfying 1852- and read as so much more impressive than the Dimension's 452. The disk score showed the WD RE4 Enterprise drive to be working well- those have 64MB cache. However, the 2D score of the massive, hot, 200W 384-bit, 1GB, 240 CUDA core GTX 285 was only 300- not as good as the Dimension's 444 from a 256MB, 16 CUDA card! The 3D of the 8400 was 261, for the Precision, it was a respectable 2208.
> It seemed impossible that the 2D of the T5400 was so poor relatively. > Something was seriously hobbling the works!
When testing the Optiplex 740, the score was 776 and like the T5400 the 2D results were mysteriously dismal- 380- but again, still better than the 300 for the GTX 285 in the Precision!
It occurred to me then that the problem must be some extreme graphic drain on the system and it took quite a while to find it. I had had an unusual problem recently in that when I installed Corel Technical Designer X5 on both the Precision and Optiplex, PhotoPaint X5 would not open multiple files. It would open the first selected file then present "I/O read error" for each of the other files.
After three re-installs of the suite- 7GB all in, numerous registry hunts and, stupidly, two 7-8 hour repair installs of Win 7, I discovered that it was my use of the Windows Classic theme in Windows 7 Ultimate that caused this error ! By trying to use the least resource-intensive setup, I had caused 20+ hours of needless, frantic effort to try and get a fairly expensive-and frequently used software to work. Corel Graphic Suite 12 was not happy on Windows 7 and running xp mode had graphic penalties- it switched to generic drivers!.
The solution to the multiple file issue, which I read on my 490th search on this subject, was to "switch from Windows Classic to an Aero Theme". I did so, choosing the one with the least number of helpful fuzzy bears.
Wow!
What I didn't realize until seeing these benchmark tests- thanks again- was that the Aero theme's transparent windows, the fading and the animations of tilting windows up and down must be seriously robbing the applications of GPU time. Window transparency seemed to be one of the hungriest components.
I sorted out a Windows 7 "Basic" theme- all solid colors- that still allowed opening multiple files in PhotoPaint X5 and ran the Precision T5400 test again. The results were astounding for this one, 5 minute change. The pleasant-sounding 1852 jumped to 2329 which I found is among the best performing single CPU T5400's and outperforming some multiple CPU and Core i7 machines. I can see too in this that DDR3 is noticeably better than DDR2 and SSD's are the future,..
Tellingly, in the new T5400 result, the 2D score jumped significantly from 300 to 583- that reads as almost doubling, 3D rose 2208 to 2320 and the CPU result changed from 4952 to 5252.
All these improvements from turning off the Aero theme!
Lesson> All the fuss I had with Corel over this issue, and all those many hours could have been saved if I had taken the time to run these tests and understand the results- a significant performance improvement for $0 that would have saved 25 hours of frustration.
Observation > Research > Analysis > Experiment
-whereas my technique of screaming at the monitor,..
The results with the Optiplex 740 were also revealing, especially that the 2.3Ghz Althlon 64 X2 4450B CPU was holding the system back. Where the T5400 Quad core Xeon was making a CPU score of 5252, the Dual core Althlon was 1441. The Windows Experience Index of 5.1 told the same story- it's the processor score lowest giving the 5.1. Again the Quadro 580 is doing well- Graphics and Gaming both 6.4.
Based on these two clear messages concerning the Optiplex CPU, I bought ($55) and installed a used Athlon 64 X2, changing the 2.30Ghz, 4450B to a 3.00Ghz 6000+, which I think also doubles the L2 cache to 2MB. (The Xeon 5460 has 12MB L2). By the way, anyone thinking of this upgrade should watch carefully for the different versions of the 6000+, the 285W PS and tiny desktop case of the Optiplex 740 desktop- and other designs apparently object to the 125W version in place of the 89W. The 89W prefix is "ADA" while the 125W can be "ADX" and possibly others. The 6000+ has been over-clocked to 3.2Ghz- supposedly reliably, and the 125W version does not have a wider over-clocking margin than the 89W. There is another 3Ghz 64 X2, the 5800, but I think that one has 1MB L2 cache instead of 2MB in the 6000+. There is also a "3.1" 6000+ , 89W and 65nm instead of 90.
Having learned the Aero theme lesson and with the 6000+ 3Ghz CPU all thermally pasted into the Optiplex 740 (my first CPU change), I ran the Passmark test again and voila, the 776 became 1113! I was quite amazed at the proportion of increase- 40%, and making it among the best performing AMD Optiplex 740's. The CPU improved 1441 to 1909 (+32%), 2D> 225 to 484 and 3D> 380 to 429. It's difficult to separate the effects of the CPU and Aero theme changes, but it seems this is, similar to the T5400, again telling about the disastrous Aero themes that the 2D graphics scores changed so significantly without any other changes.
I'm curious as to whether there is widespread similar experience with Aero themes so blatantly robbing computer performance as well as preventing Photopaint X5 from working properly. The Photopaint frustration was one the worst in my 19 years of fussing with PC's.
I was never a computer hot-rodder before, but the new imperative to do everything in 3D- even CorelDraw 2D is now Designer 3D, AutoCad became Revit and so on, makes having the best graphics system possible. - And with some careful testing and consideration, it seems really noticeable improvements can be made- and economically- to older computers, thanks to this kind of testing/analysis. I suppose new computers by definition are already running nearer the current SOTA, so the old crocks can show bigger gains,..
Sorry to go on for so very long, but I thought the discoveries made were useful and might save others' time and frustration. If you've read this far- thanks and,
Cheers,
Bambi B
Los Angeles
PS Next week, I'll see what my 1998 Dell XPS P3 750, 512MB RAM can do,..
Comment