VOGONS


First post, by Jan3Sobieski

User metadata
Rank Member
Rank
Member

Hey Everyone,

I've been doing some video card testing as of late and came to a weird conundrum you might say. I have several ATI Rage Pro Turbo cards that I have acquired and while testing them I decided to actually test each one as I figured some might be burnt or damaged. I have at least two Rage Pro Turbo AGP chipset cards, both with 8mb of vram, but it seems one on the bottom consistently scored better in everything I ran: from 3dMark99 to Quake 2, Unreal, Forsaken, Turok. If it was 0.1-0.5 fps difference, I could understand that, but for example: the top card scored 33.1fps on Turok benchmark (ran 3 times) and the bottom got 36.1 (ran 3 times again). I had even bigger fps variance in the other games, especially at the 640x480 resolution. Nothing else was changed. Used exact same drivers and testing environment (Abit BX6 rev 2.0, 1.1Ghz Pentium 3 @ 100Mhz FSB, MSI adapter, 256MB of PC100 RAM). If you look closely, the chips on both cards are exactly the same. Is it because the vram chips are different on both boards? If so, what's the difference and what can I identify it by other than physical size? Or is it because the bios chip being slightly different?

rageproturbo.th.jpg

* the bottom board has two more vram chips on the back side.

Last edited by Jan3Sobieski on 2011-03-05, 05:05. Edited 2 times in total.

Reply 5 of 17, by Old Thrashbarg

User metadata
Rank Oldbie
Rank
Oldbie

The top card is one of the cost-reduced versions, I'm not sure if it was strictly OEM-market, but that's primarily where it was aimed. Smaller board, fewer components, etc. But, more relevantly, notice the slower RAM chips... the top one has 10ns chips versus the 8ns on the bottom one. Even if the clocks are the same, the timings probably aren't. I thought the core clock would be a bit slower on the upper one too.

Also, is it sure that Powerstrip can accurately identify Rage chips?

Reply 6 of 17, by Jan3Sobieski

User metadata
Rank Member
Rank
Member
Old Thrashbarg wrote:

Also, is it sure that Powerstrip can accurately identify Rage chips?

That, I don't know. Both show, exactly: 99.76Mhz Memory clock, 74.81Mhz Engine clock.

It could be that Powerstrip just assumes all cards with this chipset run at this speed, however, I just noticed:

The top (slower) card is identified in windows as ATI Xpert 98 AGP 2x
The bottom (faster) card is identified in widows as ATI Xpert@Work AGP 2x

How can that be if they're both running the same exact chipset? The video card's bios?

Reply 8 of 17, by vlask

User metadata
Rank Member
Rank
Member

Might be memory timings. Upper one have SDR memory, other one has SGR memory. Or chip itself could be more optimized for sgr than sdr. And ofc powerstrip can be always wrong 😉

Not only mine graphics cards collection at http://www.vgamuseum.info

Reply 9 of 17, by Jan3Sobieski

User metadata
Rank Member
Rank
Member
vlask wrote:

Might be memory timings. Upper one have SDR memory, other one has SGR memory. Or chip itself could be more optimized for sgr than sdr. And ofc powerstrip can be always wrong 😉

Would this explain why the card is detected differently in windows?

Reply 10 of 17, by Tetrium

User metadata
Rank l33t++
Rank
l33t++
Jan3Sobieski wrote:
vlask wrote:

Might be memory timings. Upper one have SDR memory, other one has SGR memory. Or chip itself could be more optimized for sgr than sdr. And ofc powerstrip can be always wrong 😉

Would this explain why the card is detected differently in windows?

Possible, but not necessarily the case. It's also possible 1 is an OEM part and thus named differently.

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 11 of 17, by Old Thrashbarg

User metadata
Rank Oldbie
Rank
Oldbie

Upper one have SDR memory, other one has SGR memory.

I'm pretty sure they're both SGRAM... I don't think the Rage Pro even supported SDRAM. The Rage XL did, but I think the Rage Pro was limited to either SGRAM or WRAM.

And yes, as I said before, the top one, the Xpert98, is a low-cost version aimed primarily at OEMs. I don't see what's so hard to understand about them being named differently... it's the same general idea as, say, a Radeon 9600 vs a 9600 Pro. Just because they have the same chipset does not make them the same card.

Reply 12 of 17, by Jan3Sobieski

User metadata
Rank Member
Rank
Member
Old Thrashbarg wrote:

And yes, as I said before, the top one, the Xpert98, is a low-cost version aimed primarily at OEMs. I don't see what's so hard to understand about them being named differently... it's the same general idea as, say, a Radeon 9600 vs a 9600 Pro. Just because they have the same chipset does not make them the same card.

I understand that, I guess I'm just confused by the fact that (unless powerstrip is wrong) they both have same clock speeds. In this case the card's bios' are different, one "knowing" the card has faster ram identifies it differently even thought they have the same chipset.

Reply 13 of 17, by Old Thrashbarg

User metadata
Rank Oldbie
Rank
Oldbie

There's way more to video cards than just clockspeeds. There's a lot to do with memory timings and such going on behind the scenes. Just like with any other RAM... PC133 at CL3 is going to be slower than PC133 with CL2.

And the BIOS on the video card doesn't "know" or "identify" anything. It's pre-programmed with certain settings, and it will operate only according to the set values. The BIOS on the top card is programmed with different settings than the BIOS on the bottom card... simple as that.

Reply 15 of 17, by Jan3Sobieski

User metadata
Rank Member
Rank
Member
swaaye wrote:

How about just benching the two in say Final Reality or Quake2 and seeing what results you get.

Well, here are the results:

Top card:
Quake 2 Crusher demo benchmark
640x480 - 18.2
800x600 - 13.1
1024x768 - 8.7

Quake 2 Massive demo benchmark
640x480 - 22.4
800x600 - 15.4
1024x768 - 9.9

Final Reality score - 5.42

Bottom card:
Quake 2 Crusher demo benchmark
640x480 - 18.5
800x600 - 13.3
1024x768 - 8.9

Quake 2 Massive demo benchmark
640x480 - 22.8
800x600 - 15.8
1024x768 - 10.2

Final Reality score - 5.75

Seems like the memory timings is all that's causing this. Mystery (if you can call it that) solved.

Reply 16 of 17, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I'd like to point out two other differences. If you look at the circuit board on the bottom card, you can see the solder pads for optional memory expansion, and TV out.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium