Trident's back in the game!

[Ailuros]

Entropy,

Can I make a very bold guestimate? Since all tests at ExtremeTech were conducted in 1600x1200x32, couldn't it be that the card (for whatever possible reason) results into using only one of the 4 (1 real + 3 sharing core logic of the first) pipelines?

 

[demalion]

This is the same person and site that brought us the Anisotropic Filtering mix up. As has been mentioned earlier, this makes you wonder about things like vsync in some cases (as ATI is about to add a force vsync option to their control panel, I have a sympathetic twinge in anticipation of benchmark results based on this option being activated either accidentally, or "accidentally").

Now, it is of course possible that they have learned thoroughly from their mistakes, but with the a beta driver without even settings controls and this past history, I think it is pretty safe to say we should wait for some other sites to come up with some results.

 

[RiotSquad]

me, now using 21" sony G520 with radeon 9000.

on anybody who uses his computer mainly for "real work",
and just occasionally plays games.

Though me and others like me don't use resolutions like 1600x1200 with games,
but 1280x1024 is ok with radeon 9000 on most games.

But I agree that they should've tested with lower resolutions too.

The earlier statements about monitors was a bit "wrong"....
I mean there are people like you,but one thing I'm sure you'd agree with is that you're not as framerate hungry -or sensitive for that matter- as the people they refer too....? (e.g. powergamers...I am one myself....)
I have friends who aren't really using their PC's for work,but still find that most games are running decently....like my cousin who got a bit annoyed with how Return to castle wolfenstein was running on his Celeron 500/128 Meg/TNT2 M 64...(what _I_ would refer to as a normal PC user...surfing and also likes playing games,but nowhere near competitive gaming and are happy as long as the framerates aren't _very_ slow.....)
I personally would motice a slow down immidiately,I've been gaming since I had a P100 and played Quake at 360*240@19 FPS,which increased to "incredible" 21 FPS with a Hercules Dynamite 128 (Tseng ET6000 IIRC)...

 

[RiotSquad]

LOL @ people who had hope for Trident

Why's that....?
Imho (and I'm far from as knowledgeable as 99.9% of this forums users) I still don't know how the XP4 turns out....
I'm not expecting a miracle at their price point,I wouldn't go out and spend 100$ and think I'd come home with a high end card...however if some friend who I know are just an average user moving from say a P3 500 w/ TNT2 says build a cheap PC for me that I can do my normal stuff on the XP4 would still be a viable choice for me.('til I see a lot more of testing and that at resolutions I'd expect him to play at....)

 

[Rancidm]

The author of the article posted some lower res w/o fsaa results in their forum.

As promised, here are the kx7x32 results on 3DMark 2001SE sans FSAA. Overall, the relative performance stackup doesn't change:

3DMark 2001SE Score Fill Rate Single Fill Rate Multi Triangle Rate: 1 Light Triangle Rate: 8 Lights Nature (DX8) Vertex Shaders Pixel Shader Advanced Pixel Shader
XP4 kx7 baseline 5759 217.7 552.8 10.7 4.6 22.8 41.5 24.3 16.9
XP4 kx7 baseline (2nd run) 5745 218.1 551.9 10.7 4.6 22.8 41.5 24.3 16.9
R9000 kx7 baseline 8387 682.4 1085.4 21.6 5 41.6 79.6 113.3 78
%delta (R9000 ahead of XP4) 46.0% 212.9% 96.7% 101.9% 8.7% 82.5% 91.8% 366.3% 361.5%

I ran the XP4 twice to verify the results were repeatable.

On average, the R9000 Pro is 152% faster than the XP4 on all the tests. Looking at the overall score, R9000 Pro is still half again as fast the XP4. So yes, the XP4 does better at this resolution, but then, so does the R9000. At 1600x1200x32, the R9000 is 72%, so the XP4 was able to close the gap at 1024x768, but still trails considerably.

I agree with most of the responses in this forum. Extremetech should have been more comprehensive with thier benchmarks. I know it's their policy to bench at 1600x1200 and 1024 with AA, but that scheme obviously doesn't work optimally for all cards. Extremetech probably wants a specific set of rules to adhere to which are simple enough for the average reader to follow *and also* test hardware as concisely as possible. However, sometimes you have to "get messy" and show all the raw, complicated data to get all the facts across. Perhaps it's time for a policy change on their part.....

IMO, they shouldn't have bashed the card so hard...it's a engineering sample with beta drivers. While it probably won't turn out to be a speed-demon, anything's possible. It's a bit irresponsible to pass judgement on a product before the final silicon's even released.

 

[Tahir2]

Agreed Rancidm but remember that PR guy quoting 80% performance of GF4 Ti4600 and 70% of R9700 Pro? Those scores are a bit of a smack in the face and even with newer drivers I can't see it touching R9000 let alone the GF4 Ti 4x00's.

 

[Ailuros]

misae,

True. But the preview contained way too harsh content, which IMO shouldn't be there from any reviewer, no matter how underwhelming a test sample might perform.

I truly wonder at this point why Anand in his August preview used only 800x600 as a testing resolution. Even there the results weren't even close to a 80% of a Ti4600 claim, yet Anand was way more careful with his words, leaving a small window of hope open.

 

[Nagorak]

Maybe Trident should license ATi's AFR technology. If they put oh...say... 8 chips on the card, it will be neck in neck with the R9700!

 

[sas_simon]

me, now using 21" sony G520 with radeon 9000.

on anybody who uses his computer mainly for "real work",
and just occasionally plays games.

Yes, a work enviroment is the exception. And I wouldn't call the radeon9000 a lower then low budget card though . The cards I would have in mind would be something like the xabre and the geforce4mx 420, 440 and 460.