crenn
11-21-2007, 10:13 AM
http://www.thebestcasescenario.com/TBCS_Group/imagispire/reviews%20images/vista.png
Preview: Vista - Service Pack 1 Beta
General:
Curiosity finally got the better of me and with the permission of Jon, I'm doing a little preview of the Vista Service Pack 1 Beta. I've seen reviews saying that the SP1 beta feels faster, but none of them have really tested this (as far as I've seen).
Methods and Results:
To test whether there was any performance difference, I tested the following:
Boot-up Time
Super Pi - 1M
Super Pi - 4M
Futuremark 3DMark03
Futuremark 3DMark06
Crysis - CPU Benchmark
Crysis - GPU Benchmark
Supreme Commander
Boot up time
I measured boot up time from the time of the Boot Manager to the log in screen. Before testing, I defraged the hard drive containing Vista with a trail copy of Perfect Disk 8. I measured the boot up time 6 times with a stopwatch and then average out the results.
Before SP1 - 50.47 seconds
After SP1 - 57.86 seconds
SP1 is 7.39 seconds slower than if it wasn't installed. But this time could be different on a fresh install of Vista.
Super Pi - 1M
Super Pi relies on the CPU power and the bandwidth between the CPU and the RAM. Because of this, it can be easily be affects by another program being open. I did no 'tweaks' to improve the score for this test.
Before SP1 - 17.612 seconds
After SP1 - 17.519 seconds
Although this shows that Vista SP1 is slightly faster, there is a little problem with the result. The first time I tried the test with Vista SP1 Beta, I got almost 18 seconds. After testing it again (I tested twice) I got a faster time.
Super Pi - 4M
Similar to the test above, except Super Pi calculates 4 times the amount of places of Pi. It's the most useful test regarding modern CPUs.
Before SP1 - 1 minute 39.777 seconds
After SP1 - 1 minute 39.403 seconds
SP1 only just comes out ahead. I tested the 4M runs 3 times in SP1 to see if the same problem came up. I noticed that like the 1M run, the best time was done in the 2nd run. I'd also like to point out that when I attempted to do the 3rd run for this test, Super Pi crashed. I'm unsure of the cause.
Futuremark 3DMark03
3DMark03 tests a computer's gaming performance by stressing the CPU and GPU as a game would do.
Before SP1 - 16795 3DMarks
After SP1 - 16874 3DMarks
SP1 comes out in front by almost 100 3DMarks. The test was ran twice. The first time I tested in SP1, I got a lower score but realised that I left a few things open which affected the score a bit.
Futuremark 3DMark06
3DMark06 tests a computer's gaming performance by stressing the CPU and GPU as a game would do. This test utilises all cores of the CPU via multi-threading.
Before SP1 - 5268 3DMarks
After SP1 - 5271 3DMarks
There is a very minor performance increase here. I'm not sure I'm able to explain it. It's possible that SP1 increases performance for single threaded applications and not really multi-threaded applications.
Crysis - CPU Benchmark
On high details, this game makes almost all systems crawl. The benchmark tests the physics engine of the game. The result given is the average frames per second. The details are all set to medium.
Before SP1 - 25.56 FPS
After SP1 - 25.19 FPS
For this, SP1 actually falls behind a little. The first run for the benchmark in SP1 was a lot slower than the rest of the runs.
Crysis - GPU Benchmark
On high details, this game makes almost all systems crawl. The benchmark flies through the level testing the graphics aspects. The result given is the average frames per second. The details are all set to medium.
Before SP1 - 26.07 FPS
After SP1 - 26.04 FPS
SP1 actually falls behind again by a fraction of a second. It seems that in terms of GPU performance, there is no advantage or disadvantage for SP1.
Supreme Commander
Supreme commander puts heavy demand on the CPU with it's AI calculations. Result is the Composite and average FPS.
Before SP1 - 15972 / 30.07 FPS
After SP1 - 16031 / 30.38 FPS
SP1 runs Supreme Commander only slightly better than normal Vista.
Conclusion:
The Vista SP1 Beta is indeed faster than the normal Vista, however, the difference is marginal. Performance of crysis seemed to drop, but the difference wouldn't be noticable.
There is a couple of things I noticed which affected results a little. The boot up times for Vista, the times got as low as 35 seconds to boot up to the login screen, but it could be because I triggered the first step of getting Vista SP1 Beta. I'm not 100% sure. Another thing I noticed is the fact SP1 appears to be using more RAM than Vista! 100MB or so if I'm correct, I'm not sure why this is so as if you take a look in the screen shots, you'll see what I mean.
It's possible that more improvements will surface when Vista SP1 is officially released. I'm wondering if there is debugging tools running in the background that could be affecting the performance slightly. I'll review SP1 when it's released to see how it stands up. If you would like the raw data I collected during these tests, PM me and I'll organise to email them to you.
Additional Comments:
I also played a bit of the Crysis Demo to see what the game 'felt' like. And I noticed that playing it in Vista SP1 compared to Vista, there was a very noticeable difference. In Vista, the mouse seemed to be lagging, but in SP1, I had no problems.
I apologize for not having pretty graphs for this review.
Preview: Vista - Service Pack 1 Beta
General:
Curiosity finally got the better of me and with the permission of Jon, I'm doing a little preview of the Vista Service Pack 1 Beta. I've seen reviews saying that the SP1 beta feels faster, but none of them have really tested this (as far as I've seen).
Methods and Results:
To test whether there was any performance difference, I tested the following:
Boot-up Time
Super Pi - 1M
Super Pi - 4M
Futuremark 3DMark03
Futuremark 3DMark06
Crysis - CPU Benchmark
Crysis - GPU Benchmark
Supreme Commander
Boot up time
I measured boot up time from the time of the Boot Manager to the log in screen. Before testing, I defraged the hard drive containing Vista with a trail copy of Perfect Disk 8. I measured the boot up time 6 times with a stopwatch and then average out the results.
Before SP1 - 50.47 seconds
After SP1 - 57.86 seconds
SP1 is 7.39 seconds slower than if it wasn't installed. But this time could be different on a fresh install of Vista.
Super Pi - 1M
Super Pi relies on the CPU power and the bandwidth between the CPU and the RAM. Because of this, it can be easily be affects by another program being open. I did no 'tweaks' to improve the score for this test.
Before SP1 - 17.612 seconds
After SP1 - 17.519 seconds
Although this shows that Vista SP1 is slightly faster, there is a little problem with the result. The first time I tried the test with Vista SP1 Beta, I got almost 18 seconds. After testing it again (I tested twice) I got a faster time.
Super Pi - 4M
Similar to the test above, except Super Pi calculates 4 times the amount of places of Pi. It's the most useful test regarding modern CPUs.
Before SP1 - 1 minute 39.777 seconds
After SP1 - 1 minute 39.403 seconds
SP1 only just comes out ahead. I tested the 4M runs 3 times in SP1 to see if the same problem came up. I noticed that like the 1M run, the best time was done in the 2nd run. I'd also like to point out that when I attempted to do the 3rd run for this test, Super Pi crashed. I'm unsure of the cause.
Futuremark 3DMark03
3DMark03 tests a computer's gaming performance by stressing the CPU and GPU as a game would do.
Before SP1 - 16795 3DMarks
After SP1 - 16874 3DMarks
SP1 comes out in front by almost 100 3DMarks. The test was ran twice. The first time I tested in SP1, I got a lower score but realised that I left a few things open which affected the score a bit.
Futuremark 3DMark06
3DMark06 tests a computer's gaming performance by stressing the CPU and GPU as a game would do. This test utilises all cores of the CPU via multi-threading.
Before SP1 - 5268 3DMarks
After SP1 - 5271 3DMarks
There is a very minor performance increase here. I'm not sure I'm able to explain it. It's possible that SP1 increases performance for single threaded applications and not really multi-threaded applications.
Crysis - CPU Benchmark
On high details, this game makes almost all systems crawl. The benchmark tests the physics engine of the game. The result given is the average frames per second. The details are all set to medium.
Before SP1 - 25.56 FPS
After SP1 - 25.19 FPS
For this, SP1 actually falls behind a little. The first run for the benchmark in SP1 was a lot slower than the rest of the runs.
Crysis - GPU Benchmark
On high details, this game makes almost all systems crawl. The benchmark flies through the level testing the graphics aspects. The result given is the average frames per second. The details are all set to medium.
Before SP1 - 26.07 FPS
After SP1 - 26.04 FPS
SP1 actually falls behind again by a fraction of a second. It seems that in terms of GPU performance, there is no advantage or disadvantage for SP1.
Supreme Commander
Supreme commander puts heavy demand on the CPU with it's AI calculations. Result is the Composite and average FPS.
Before SP1 - 15972 / 30.07 FPS
After SP1 - 16031 / 30.38 FPS
SP1 runs Supreme Commander only slightly better than normal Vista.
Conclusion:
The Vista SP1 Beta is indeed faster than the normal Vista, however, the difference is marginal. Performance of crysis seemed to drop, but the difference wouldn't be noticable.
There is a couple of things I noticed which affected results a little. The boot up times for Vista, the times got as low as 35 seconds to boot up to the login screen, but it could be because I triggered the first step of getting Vista SP1 Beta. I'm not 100% sure. Another thing I noticed is the fact SP1 appears to be using more RAM than Vista! 100MB or so if I'm correct, I'm not sure why this is so as if you take a look in the screen shots, you'll see what I mean.
It's possible that more improvements will surface when Vista SP1 is officially released. I'm wondering if there is debugging tools running in the background that could be affecting the performance slightly. I'll review SP1 when it's released to see how it stands up. If you would like the raw data I collected during these tests, PM me and I'll organise to email them to you.
Additional Comments:
I also played a bit of the Crysis Demo to see what the game 'felt' like. And I noticed that playing it in Vista SP1 compared to Vista, there was a very noticeable difference. In Vista, the mouse seemed to be lagging, but in SP1, I had no problems.
I apologize for not having pretty graphs for this review.