Updated video drivers now 2d test very slow.

[radman3d]

Ran test and all went well, but PCPitstop told me to update my video card drivers which I did, to the version it posted. Now the 2d test is very slow, line drawing in particular is very slow. I have an FX5200 video card. What could be the problem? Could it be a setting and if so which setting?

[IntelGuy]

 

Usually with the cards that have been available for a while the issue is that the newer driver has decreased 2D support. The first thing to try would be to roll back the driver and see if the issue is resolved.

 

However, if everthing on the computer is working well and it is only the OverDrive 2D score, I would suggest ignoring the low score.

[dmart7]

I think there is a bit more at issue here. I have had the same problem on an Nvidia-based 9600GT for some months. A scan on the internet revealed that a large number of people have these odd low 2d readouts on Pitstop (with the ellipse line drawing test appearing to be the problem). Other test programmes revealed no 2d problem on offending machines, contradicting the PcPitstop results. Updating drivers etc had no effect. Rightly or wrongly, I therefore assumed that PcPitstop had a bug or incompatibility with some cards or drivers, and that this would eventually be fixed. If so, it hasn't.

 

For the record, my 3d video benchmark is 4706fps (top 10%) whiler 2d video is 96 MP/S (bottom 30%). I've never detected low 2d performance in normal use.

 

Whatever the cause, it is card related. I have other PCs which do not suffer this problem on PC Pitstop.

 

Grateful for any further comments.

[IntelGuy]

As far as I have seen, that situation only occurs when using newer drivers on older cards. I would suggest rolling back the driver to check if the issue is resolved.

[dmart7]

This is a new PC (three months) with equally new graphics card. Since I have no reason to believe that the low 2d mark is true, or that I actually have an "issue", then it must be a bug in PcPitstop tests, which cannot cope with certain cards or their drivers.

 

If true, I would have thought someone might want to try to resolve it.

[ginahoy]

I spent several hours over the past week troubleshooting a similar problem. In my case, the card is almost two years old (7600GS AGP). Although my card's 3D performance was nominal, the 2D score was 10.6 MP/s compared to 118 MP/s typical for my card! PCP advised updating the drivers. That made sense. I hadn't updated my drivers in more than 18 months. In hindsight, I wish I'd never visited PCP.

 

As it turns out, due to a problem with Nvidia's most recent drivers (Code 10 error on some cards), I had to revert to the original drivers that shipped with my card. I spent some time troubleshooting this with Sparkle, tech support, which got me nowhere.

 

For grins, I ran the PCP tests again with the old drivers. I was surprised to see my 2D score jump to 117 MP/s, pretty much nominal for my card according to PCP. The ellipse draw test just flew by.

 

At first, this result seems to support IntelGuy's comment (re: new vs. old drivers on older cards). However, I saw *no* improvement in daily use of 2D graphics applications. I think a 10-fold increase as indicated by the test would certainly be evident. I began to suspect a problem with the PCP 2D test.

 

I had some time to kill this weekend and tried several intermediate drivers sets available on Nvidia's website. As it turns out, the 10-fold drop in score on PCP's 2D test began with drivers released in the 2nd half of 2007 (169.21 for non-quadro cards).

 

Two days ago, Nvidia release a new set of certified drivers for my card, which solved the Code 10 error. I just installed these and ran the PCP test suite again, but got the same low 2D score.

 

Wanting an independent confirmation of PCP test results, I did some research and discovered Performance Test 7.0 by PassMark. As it turns out, my scores on the PT7 2D test are relatively flat across old and new drivers. If anything, the latest driver set turned in slightly higher results.

 

Based on my experience, and on comments from dmart7 and others, I think there's a specific compatibility issue here with PCP's 2D test. Something is causing the 2D test to perform poorly with newer Nvidia drivers, especially the ellipse draw. I don't buy the argument that driver evolution is responsible for impaired 2D performance, in my case, by a factor of 10.

 

I'm sure there's a logical explanation for this. Perhaps someone with a deep understanding of the inner workings of these tests would care to comment?

Edited May 9, 2009 by ginahoy

[IntelGuy]

The developer that wrote the 2D test has been notified of your concerns.

 

Performance Test by Passmark checks 2D vectors, image filtering and rendering, and fonts and texts. There is no line-drawing test in Performance Test. It becomes a comparison of the testing methodologies, not of the resuts. Passmark is not a valid comparison as you end up comparing apples to oranges.

 

I do not follow the logic of your conclusion. As you yourself pointed out, Nvidia drivers prior to July of 2007 score well and drivers after that time score poorly. What changed was the driver, not the 2D test.

[dmart7]

I have not attended this thread for many months - lent the offending PC to my son - but now have it back. The 2d PCP tests remain as illogical as ever on my 9600GT and Ginahoy's analysis reinforces my own. All other benchmark tests give these cards a normal 2d rating.

 

The fact that it is suggested that the problem only occurs in XP (and is specifically a driver problem only in XP) is interesting, and contradicts the assertion that the cards are intrinsically poor at 2d..

 

The interest is more than just academic. It is not that I just want a higher "score". I need to know whether there is something wrong with my 2d graphics, especially as I expect to use this more than the 3d (eg. including video editing)

 

My question:

 

Will it definitely improve if I upgrade to Windows 7? A lot?

 

Thanks

Edited January 27, 2010 by dmart7

[IntelGuy]

Your mileage may vary.

 

My GTX 295 went from a 2D score of 158 to a score of 578 (at a lower clock speed as well) by upgrading to Windows 7:

 

XP: http://www.pcpitstop.com/betapit/sec.asp?c...;report=Summary

 

Win7: http://www.pcpitstop.com/betapit/sec.asp?c...;report=Summary

[dmart7]

Once again, I come back to this issue after a gap. I now have Windows 7 installed and a different graphics card - ATI Radeon HD5450. Everything is fully updated and runs perfectly with good Pitstop scores. However, there is curious a 2D issue. If I have Aero turned off (in the hope that things will be faster!) the 2D scores are low (c. 115). If I have Aero on, the 2D speeds become much faster (c. 250).

 

Obviously, I shall keep Aero on. Interestingly, I can turn Aero peek off, but if I disable Desktop Composition, the scores plummet to 115 again.

 

My question: is this counter-intuitive behaviour a function (or fault?) of the PC Pitstop tests or of the graphics card drivers?