Pixel shader position switcheroo in 3dMark05...

Subtestube · 09-30-04, 06:28 PM

I'm just generally wondering how it is that the typical situation we've seen before has been switched in 3dMark05. I'm not complaining - I'm really interested in what it is about the card architecture that's providing the X800 with better Pixel shader performance than the 6800, when everywhere else we've seen the 6800 series seems to take the lead in fragment shader performance? It seems a little odd to me.

[Note: I'm talking about the strictly Pixel shader test - not the in games]. So... any ideas? Where is it in the X800 architecture that it's getting such radical gains in PS performance?

jbirney · 10-01-04, 11:41 AM

Subtestube,

thats a good point. However synthetics are forward looking benchmarks. The level, ammount and style of shaders they use are "ahead" of how games will/may use them. Trying to compare current day perfromance to 3Dmark is not a meaning full comparision. For example one of the scenes in 3dmark2k5 hits the 4 million polygon mark. No game today comes close to that. So we need to spend more time understanding how 3dmark2k5 works and give NV/ATI time to still tweak their hardware before we can know for certian if there is a real limitation that cause one to be faster..or something else.

subbo · 10-01-04, 04:47 PM

To answer (or guess) your question about the superior performance, I'd say its down to the raw number crunching power of the core clock. The 8000+ score was done with 850Mhz+ which is twice what an Ultra can offer, and so is (almost) the score.

rohit · 10-03-04, 05:29 AM

I feel the same,
Ive seen only overcloked Ultras and GT's get 5K+ scores, n some x800 pros scorin more than an ultra runnin at stock speeds. I guess this has to be related with the core frequency. Higher the core frequency.. higher the score in ,05'.
Even the highest scoring XTPE was cloked at 806MHz core.. whereas the 2 Sli Ultras wer cloked at stock 400 each...
I really dont understand wheather the results are PS bound or FillRate Bound.

Enlighten me if i am wrong!

Btw i scored 3915 with mah MSI GT at stock speed. usin 61.77 drivers
With my rig not even attaining 4k is sad, bad & is drivin me mad.
I dont bleve in oc'ing of mah card coz all the games are runnin absolutely great on stock speeds of my card.
Ocing of mah processor didnt make shi* difference.
it wuldnt hav made any diff anyways, coz ive read the P4 is defeating the athlon64 ... lol thats more than funny.

NEODARK · 10-05-04, 10:53 AM

its the drivers you are using, get 66.70's or 66.81's they will put you around 4500+ with no tweaks

09-30-04, 06:28 PM	#1
Subtestube Anisymbolic Join Date: Aug 2004 Location: Wellington, New Zealand Posts: 1,365	Pixel shader position switcheroo in 3dMark05... I'm just generally wondering how it is that the typical situation we've seen before has been switched in 3dMark05. I'm not complaining - I'm really interested in what it is about the card architecture that's providing the X800 with better Pixel shader performance than the 6800, when everywhere else we've seen the 6800 series seems to take the lead in fragment shader performance? It seems a little odd to me. [Note: I'm talking about the strictly Pixel shader test - not the in games]. So... any ideas? Where is it in the X800 architecture that it's getting such radical gains in PS performance? __________________ Dr Possible: Core 2 Duo E6400 on Gigabyte GA-965P-DS4. Galaxy GeForce 7600GT. 2GB Corsair XMS 2 DDR2-6400 RAM (CL5). ATi Theatre 550 Pro. Windows XP MCE. All stored in Piano black Antec Sonata II, with a broken door. Mobile: ASUS M2400N, Pentium M 1.5 GHz. 512 MB DDR RAM. Intel EXTREME graphics. Windows XP SP 2 / Ubuntu 5.10. Ridiculous DOES not have an 'e' in it. It comes from "ridicule" and has less than nothing to do with the colour red.

10-01-04, 11:41 AM	#2
jbirney Registered User Join Date: Jul 2002 Posts: 1,430	Re: Pixel shader position switcheroo in 3dMark05... Subtestube, thats a good point. However synthetics are forward looking benchmarks. The level, ammount and style of shaders they use are "ahead" of how games will/may use them. Trying to compare current day perfromance to 3Dmark is not a meaning full comparision. For example one of the scenes in 3dmark2k5 hits the 4 million polygon mark. No game today comes close to that. So we need to spend more time understanding how 3dmark2k5 works and give NV/ATI time to still tweak their hardware before we can know for certian if there is a real limitation that cause one to be faster..or something else.

10-01-04, 04:47 PM	#3
subbo Registered User Join Date: May 2003 Posts: 452	Re: Pixel shader position switcheroo in 3dMark05... To answer (or guess) your question about the superior performance, I'd say its down to the raw number crunching power of the core clock. The 8000+ score was done with 850Mhz+ which is twice what an Ultra can offer, and so is (almost) the score. __________________ A64 3200+, MSI NX6800GT, Asus K8V SE D, NEW 2GB DDR400 (BF2 likey), 200GB WD Caviar, Nexus 4090 400W silent, ATC220B, Zalman Reserator 1, Innovatek CoolMatic NV40, NEC 3500AG, CM AerogateII black, Smart Drive 2002C \| 20,1" 15ms TFT 1600x1200 Formac Gallery 2010 \| Temps: idle CPU 45c GPU 55c, 1h+ gaming CPU 55c GPU 67c

10-03-04, 05:29 AM	#4
rohit Awaiting WGF 2.0 Join Date: Jul 2004 Posts: 1,174	Re: Pixel shader position switcheroo in 3dMark05... I feel the same, Ive seen only overcloked Ultras and GT's get 5K+ scores, n some x800 pros scorin more than an ultra runnin at stock speeds. I guess this has to be related with the core frequency. Higher the core frequency.. higher the score in ,05'. Even the highest scoring XTPE was cloked at 806MHz core.. whereas the 2 Sli Ultras wer cloked at stock 400 each... I really dont understand wheather the results are PS bound or FillRate Bound. Enlighten me if i am wrong! Btw i scored 3915 with mah MSI GT at stock speed. usin 61.77 drivers With my rig not even attaining 4k is sad, bad & is drivin me mad. I dont bleve in oc'ing of mah card coz all the games are runnin absolutely great on stock speeds of my card. Ocing of mah processor didnt make shi* difference. it wuldnt hav made any diff anyways, coz ive read the P4 is defeating the athlon64 ... lol thats more than funny. __________________ \|C.P.U : s939 AMD Athlon64 3200+ (newcastle) \| \|MoBo : ASUS A8V Deluxe \|\|RAM: [512 x 2]MB Transcend DDR400 (pc3200) \| \|V.G.A : LeadTek a400GT TDH (350/1000) [GeForce 6800GT] \| \|Sound: RealTek Integrated. \| \|P.S.U : PowerSafe 450w \| \|MISC.: Logitech Cordless DUO(MX700 + Wireless Keyboard)\|\|Lite-On DVD CD-RW Combo 52x16x52 \|3dMark'05: 4893 [@ 350/1000]

10-05-04, 10:53 AM	#5
NEODARK Haloed out Join Date: Oct 2004 Posts: 21	Re: Pixel shader position switcheroo in 3dMark05... its the drivers you are using, get 66.70's or 66.81's they will put you around 4500+ with no tweaks __________________ A64 3500+ \|\| 1gig OCZ 3700EB Platinum \|\| MSI K8N Neo2 Plat \|\| BFG 6800 OC \|\| Audigy2 \|\| Tsunami Dream VA3000

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Pixel shader version 1.4 supported?	JohnsonLKD	NVIDIA GeForce 7, 8, And 9 Series	19	08-20-02 04:04 AM