Different driver path for X800 XT and pro?

[Broken Hope]

I have a VIVO pro with 16 pipes unlocked, and when flashed with an XT bios I get around 150 fps in the pixel shader 2 test on 3dmark03, yet the same card flashed with a pro bios, using the pro drivers, but at the same clock speed only scores 120 frames in the exact same test.

What happened to the other 30 fps? considering the same number of pipes and the same clock speed, the only difference is the bios and I assume the driver loaded.

The odd thing is though, that the overall score is the same, so whatever the problem is, it doesn't seem to be affecting the game tests.

 

[Rys]

Surely the PRO BIOS disables a quad pipe? You're losing ~25% of performance. One quad pipe disabled is 25% of the processing power.

Coincidence?

Rys

 

[MasterBaiter]

Could you verify how many pipes you have for us after the flash? You should be able to use ATITool: http://atitool.ocfaq.com/

My Gigabyte x800 Pro Vivo had 16 pipes on by default which was a pleasant surprise.

 

[Dave Baumann]

I would imagine that, ideally, you would want different memory mappings for 3 quad and 4 quad parts as well.

 

[Broken Hope]

Here's my compare link, the fill rate show's that I can't be running on 12 pipes and this picture from ATI tool shows the 16 pipes

http://service.futuremark.com/compare?2k3=2749751

 

[Bouncing Zabaglione Bros.]

Well I've found almost no difference in 3Dmark scores going from 12 pipes to 16 pipes on a X800 Pro. The scores are also very close when using a XTPE BIOS clocked down to Pro levels, ie, the difference only seems to be due to core/memory clocks between Pro and XTPE BIOSes.

 

[Broken Hope]

You're right the overall score is the same, but are your pixel shader 2 tests the same on the pro and XT bios?

 

[Bouncing Zabaglione Bros.]

You're right the overall score is the same, but are your pixel shader 2 tests the same on the pro and XT bios?

I'm getting the same results as you - 120 on the Pro BIOS, 150 on the ATI XTPE BIOS. What I'd really like to try is one of the newer Sapphire XTPE BIOSes.

 

[Broken Hope]

You're right the overall score is the same, but are your pixel shader 2 tests the same on the pro and XT bios?

I'm getting the same results as you - 120 on the Pro BIOS, 150 on the ATI XTPE BIOS. What I'd really like to try is one of the newer Sapphire XTPE BIOSes.

I tried an XTPE bios extracted from a retail Sapphire card with the same results.

 

[Bouncing Zabaglione Bros.]

You're right the overall score is the same, but are your pixel shader 2 tests the same on the pro and XT bios?

I'm getting the same results as you - 120 on the Pro BIOS, 150 on the ATI XTPE BIOS. What I'd really like to try is one of the newer Sapphire XTPE BIOSes.

I tried an XTPE bios extracted from a retail Sapphire card with the same results.

Well it looks to be the case that unlocking the extra quad isn't enough. Without an XTPE bios, the cards either ignores the quad, or doesn't have the memory timings/info/whatever to utilise the extra pipes.

Actually, what I'd really like is a XTPE Vivo BIOS that utilises the extra quad, but at the standard Pro speeds. Then you wouldn't have to downclock the XTPE bios to use the extra pipes.

 

[Broken Hope]

You're right the overall score is the same, but are your pixel shader 2 tests the same on the pro and XT bios?

I'm getting the same results as you - 120 on the Pro BIOS, 150 on the ATI XTPE BIOS. What I'd really like to try is one of the newer Sapphire XTPE BIOSes.

I tried an XTPE bios extracted from a retail Sapphire card with the same results.

Well it looks to be the case that unlocking the extra quad isn't enough. Without an XTPE bios, the cards either ignores the quad, or doesn't have the memory timings/info/whatever to utilise the extra pipes.

Actually, what I'd really like is a XTPE Vivo BIOS that utilises the extra quad, but at the standard Pro speeds. Then you wouldn't have to downclock the XTPE bios to use the extra pipes.

Oh I'm sure the pipes are working, I get much higher fill rate in the fill rate tests, I gained around 1000 points in 3dmark over a 12 pipe pro at XT speeds and even at pro clocks with the extra pipes shows improvement over a 12 pipe pro especially in GT 2 on 3dmark03

 

[Bouncing Zabaglione Bros.]

Oh I'm sure the pipes are working, I get much higher fill rate in the fill rate tests, I gained around 1000 points in 3dmark over a 12 pipe pro at XT speeds and even at pro clocks with the extra pipes shows improvement over a 12 pipe pro especially in GT 2 on 3dmark03

So how do you explain the significant difference in the PS2.0 tests? There's something in the XT BIOS that is using those extra pipes in a way that the Pro BIOS doesn't.

 

[Broken Hope]

Oh I'm sure the pipes are working, I get much higher fill rate in the fill rate tests, I gained around 1000 points in 3dmark over a 12 pipe pro at XT speeds and even at pro clocks with the extra pipes shows improvement over a 12 pipe pro especially in GT 2 on 3dmark03

So how do you explain the significant difference in the PS2.0 tests? There's something in the XT BIOS that is using those extra pipes in a way that the Pro BIOS doesn't.

I'm not sure why the big difference which is why I made this thread, maybe it's as Dave said and that there is different mapping for the extra quad? Or it could be a different driver path, have you tried unchecking the show only compatible hardware box in a manual driver install and forcing the XT driver onto the pro?

What I don't understand though is that the overall score for the game tests are the same for both bios's only the ps2.0 test shows a difference.

I'm hoping that maybe ATI will release an AGP version of the normal XT which is clocked at 500/500

 

[Star_Hunter]

I am wondering.... Do the 9800, 9800 Pro, and 9800 XT all use the same path or do they do the same thing as seen here? Also some people have said that since Ati is now looking at the X800 series for performance increases that this would lower the performance of the R3x0 series, if they are all running on differnt paths shouln't one not really effect the other and therefor be able to increase both rather then tweaking the X800 to be faster causes the R3x0 to be slower due to the differences?