• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

why is the 2900xt slower than the 8800gtx

Darkmind

New Member
Joined
May 23, 2005
Messages
159 (0.02/day)
Processor Intel Pentium Dual-Core E6600 @ 3GHz (333x9)
Motherboard Gigabyte GA-P35-DS3L
Cooling CM GeminII S
Memory G.SKILL 2x2GB @ DDR2 1000 @ 5-5-5-15
Video Card(s) XFX ATI Radeon 5870 1GB
Storage Seagate Barracuda 7200.10 320GB
Display(s) Samsung 2220ex 21.5in
Case Antec 900 w/ 3 120mm intake fans, 1 120mm rear exhaust fan, 1 200mm top exhaust fan
Audio Device(s) SB Audigy 2ZS w/ Logitech x-530 5.1 speakers
Power Supply Antec Trupower 750w
Software Windows Vista Ultimate 32-bit
Spec wise it kicks ass except the amount of ROPs. Is it the amount of ROP's that made it so it wouldn't take the performance crown? 16 vs 24. But it has 320 shader processors (clocked at 742mhz), while the 8800gtx is 128 shader processors at 1.35ghz.

I guess I must've answered my own question? Hmm
 
V

v-zero

Guest
It's not the ROPs, it's the design. Those "320 shaders" are deceptive - most are not fully developed shaders, and so do not have anywhere near the potency of each shader in the G80 based cards. As such, the technically superior G80 will always win.
 
Joined
Aug 20, 2007
Messages
20,787 (3.41/day)
System Name Pioneer
Processor Ryzen R9 7950X
Motherboard GIGABYTE Aorus Elite X670 AX
Cooling Noctua NH-D15 + A whole lotta Sunon and Corsair Maglev blower fans...
Memory 64GB (4x 16GB) G.Skill Flare X5 @ DDR5-6000 CL30
Video Card(s) XFX RX 7900 XTX Speedster Merc 310
Storage 2x Crucial P5 Plus 2TB PCIe 4.0 NVMe SSDs
Display(s) 55" LG 55" B9 OLED 4K Display
Case Thermaltake Core X31
Audio Device(s) TOSLINK->Schiit Modi MB->Asgard 2 DAC Amp->AKG Pro K712 Headphones or HDMI->B9 OLED
Power Supply FSP Hydro Ti Pro 850W
Mouse Logitech G305 Lightspeed Wireless
Keyboard WASD Code v3 with Cherry Green keyswitches + PBT DS keycaps
Software Gentoo Linux x64
I always thought it wasn't so much that the shader were less developed as it was that ATI went with a less efficient super-scalar architechture that their dispatcher isn't so efficient at handling. That and the slower clock speeds of the shaders does put them at a disadvantage.

If you want to go by paper, technically the ATI piece of hardware is "technically superior" in that it has more transistors and thus more features (tesselator and all that). But those features aren't part of the DX10 spec, and as such probably won't be used in many games. Also, this superiority in features does nothing when it comes to FPS, where NVIDIA consistently wins for the reasons I listed above.

Maybe I misread a few things though, that's just how I understood it from the article at anandtech.
 

Matsu

New Member
Joined
Sep 26, 2007
Messages
5 (0.00/day)
AMD explains Radeon HD 2900XT's poor AA performance 1:16PM, Monday 14th May 2007

The R600 is finally here, and in keeping with its mysteriously long gestation, in at least in its first incarnation as the HD 2900XT, AMD's new GPU still poses a lot of questions. One of the things we noticed during our in-depth testing of the card is that compared to its principle rival, the Nvidia GeForce 8800 GTS 640MB, the HD 2900XT performs poorly in many games when anti-aliasing is enabled.

In F.E.A.R., at 1,600 x 1,200, with AA and AF disabled, the HD 2900XT easily outstripped the 640MB 8800 GTS, delivering a minimum that was 23fps higher than the latter's. However, with 4x AA, the HD 2900XT's minimum framerate dived from 82fps to 21fps, while the 640MB 8800 GTS produced a minimum of 30fps. Adding 4x AA results in a 74% drop for the Radeon, compared to only a 49% drop for the GeForce.

The Radeon's framerates suffer disproportionately with anisotropic filtering, too. Again testing in F.E.A.R. at 1,600 x 1,200, we saw the HD 2900XT's minimum FPS drop by 10 per cent with 16x anisotropic enabled, compared to 3 per cent for the GTS, although the HD 2900XT still had a faster average. It was a slightly different result at 2,560 x 1,600, as the HD 2900XT's massive bandwidth gave it a boost, although adding 16x AF still had more impact than it did on the 640MB GTS.

As most gamers will want AA and AF enabled in games, the HD 2900XT's poor performance with these processing options enabled is a serious problem for the card and ATi. We asked ATi to comment on this surprising result and the company revealed that the HD 2000-series architecture has been optimised for what it calls 'shader-based AA'. Some games, including S.T.A.L.K.E.R., already use shader-based AA, although in our tests the 640MB 8800 GTS proved to be faster than the HD 2900XT.

We asked Richard Huddy, Worldwide Developer Relations Manager of AMD's Graphics Products Group, to go into more detail about why the Radeon HD 2000-series architecture has been optimised for shader-based AA rather than traditional multi-sample AA. He told us that 'with the most recent generations of games we've seen an emphasis on shader complexity (mostly more maths) with less of the overall processing time spent on the final part of the rendering process which is "the AA resolve". The resolve still needs to happen, but it's becoming a smaller and smaller part of the overall load. Add to that the fact that HDR rendering requires a non-linear AA resolve and you can see that the old fashioned linear AA resolve hardware is becoming less and less significant.' Huddy also explained that traditional AA 'doesn't work correctly [in games with] HDR because pixel brightness is non-linear in HDR rendering.'

While many reviews of the HD 2900XT have made unflattering comparisons between it and Nvidia's GeForce 8800-series, Huddy was upbeat about AMD's new chip. 'Even at high resolutions, geometry aliasing is a growing problem that can only really be addressed by shader-based anti-aliasing. You'll see that there is a trend of reducing importance for the standard linear AA resolve operation, and growing importance for custom resolves and shader-based AA. For all these reasons we've focused our hardware efforts on shader horsepower rather than the older fixed-function operations. That's why we have so much more pure floating point horsepower in the HD 2900XT GPU than NVIDIA has in its 8800 cards... There's more value in a future-proof design such as ours because it focuses on problems of increasing importance, rather than on problems of diminishing importance."
.

If you are talking about AA performance then the R600 isn't supposed to be good at Multi sampling AA ( I think that is what its called). It is supposed to be good at Shader AA revolve. I think it works by using the shaders to do the AA instead.
 

KennyT772

New Member
Joined
Dec 21, 2005
Messages
3,572 (0.53/day)
System Name Raptor
Processor Intel E8400 Wolfdale @3600
Motherboard Asus P5Q-Pro LGA775
Cooling Zalman CNPS9700
Memory 1024mbx2 Crucial Ballistix DDR800
Video Card(s) XFX 9600GT
Storage Seagate 7200.11 500GB
Display(s) Acer AL2216Wbd and Acer AL1717
Case Gigabyte 3DAurora Black
Audio Device(s) Creative X-Fi Extreme Gamer Faitality
Power Supply OCZ GameXStream 500w
The 2900 is slower one because of the superscalar design - parallelism is key. If the 2900 had half the shaders but running twice as fast you would see performance increase by 20%.

Also the 2900 is a brand new card compared to the 8800. Ati has simply not had the same amount of time to optimize and improve the drivers as nVidia has. I'm not just talking about making the game run as fast as possible, but also various quality settings, various system configurations, etc.
 
Joined
Aug 20, 2007
Messages
20,787 (3.41/day)
System Name Pioneer
Processor Ryzen R9 7950X
Motherboard GIGABYTE Aorus Elite X670 AX
Cooling Noctua NH-D15 + A whole lotta Sunon and Corsair Maglev blower fans...
Memory 64GB (4x 16GB) G.Skill Flare X5 @ DDR5-6000 CL30
Video Card(s) XFX RX 7900 XTX Speedster Merc 310
Storage 2x Crucial P5 Plus 2TB PCIe 4.0 NVMe SSDs
Display(s) 55" LG 55" B9 OLED 4K Display
Case Thermaltake Core X31
Audio Device(s) TOSLINK->Schiit Modi MB->Asgard 2 DAC Amp->AKG Pro K712 Headphones or HDMI->B9 OLED
Power Supply FSP Hydro Ti Pro 850W
Mouse Logitech G305 Lightspeed Wireless
Keyboard WASD Code v3 with Cherry Green keyswitches + PBT DS keycaps
Software Gentoo Linux x64
Ah, I forgot about the shaders. Honestly I don't use AA at anything over 1600x1200 because I don't notice it so that doesn't effect me. But yeah, it doesn't have any MSAA hardware, it just uses shaders, so it'll perform far better in none AA benchmarks.
 
Joined
Dec 5, 2006
Messages
7,704 (1.21/day)
System Name Back to Blue
Processor i9 14900k
Motherboard Asrock Z790 Nova
Cooling Corsair H150i Elite
Memory 64GB Corsair Dominator DDR5-6400 @ 6600
Video Card(s) EVGA RTX 3090 Ultra FTW3
Storage 4TB WD 850x NVME, 4TB WD Black, 10TB Seagate Barracuda Pro
Display(s) 1x Samsung Odyssey G7 Neo and 1x Dell u2518d
Case Lian Li o11 DXL w/custom vented front panel
Audio Device(s) Focusrite Saffire PRO 14 -> DBX DriveRack PA+ -> Mackie MR8 and MR10 / Senn PX38X -> SB AE-5 Plus
Power Supply Corsair RM1000i
Mouse Logitech G502x
Keyboard Corsair K95 Platinum
Software Windows 11 x64 Pro
Benchmark Scores 31k multicore Cinebench - CPU limited 125w
Also remember it depends on what the game itself was tweaked for...

The 8800 has been out long enough for quite a few new games to be built to perform well on it, while the 2900 was not in existence. Thats another piece that helps..
 
Joined
May 6, 2005
Messages
2,792 (0.40/day)
Location
Tre, Suomi Finland
System Name Ladpot ◦◦◦ Desktop
Processor R7 5800H ◦◦◦ i7 4770K, watercooled
Motherboard HP 88D2 ◦◦◦ Asus Z87-C2 Maximus VI Formula
Cooling Mixed gases ◦◦◦ Fuzion V1, MCW60/R2, DDC1/DDCT-01s top, PA120.3, EK200, D12SL-12, liq.metal TIM
Memory 2× 8GB DDR4-3200 ◦◦◦ 2× 8GB Crucial Ballistix Tactical LP DDR3-1600
Video Card(s) RTX 3070 ◦◦◦ heaps of dead GPUs in the garage
Storage Samsung 980 PRO 2TB ◦◦◦ Samsung 840Pro 256@178GB + 4× WD Red 2TB in RAID10 + LaCie Blade Runner 4TB
Display(s) HP ZR30w 30" 2560×1600 (WQXGA) H2-IPS
Case Lian Li PC-A16B
Audio Device(s) Onboard
Power Supply Corsair AX860i
Mouse Logitech MX Master 2S / Contour RollerMouse Red+
Keyboard Logitech Elite Keyboard from 2006 / Contour Balance Keyboard / Logitech diNovo Edge
Software W11 x64 ◦◦◦ W10 x64
Benchmark Scores It does boot up? I think.
Basically the shaders on R600 can't calculate as complex functions as those on G80 and on top of that they do less work per clock.
 
V

v-zero

Guest
A said earlier, most of the shaders in the R600 design are very limited, with a few powerful shaders (more similar to the 8800's) carrying much of the load. Also, the lack of proper AA support causes massive problems.
 

Matsu

New Member
Joined
Sep 26, 2007
Messages
5 (0.00/day)
Games that actually use shaders to do AA the ATI cards won't look as bad.
 
Joined
Nov 16, 2006
Messages
2,724 (0.43/day)
Location
Blighty
Processor R7 5800x3D
Motherboard MSI x570 Tomahawk
Cooling XSPC Raystorm Edge,EK QS P420M,EK D5pwm Revo Res
Memory 32gb Corsair Vengeance RT 3600 cl16
Video Card(s) Zotac 3070ti Amp Extreme
Storage Samsung 980pro 1tb x2
Display(s) MSI MPG321QRF QD
Case Corsair 7000D
Power Supply EVGA 1000 P2
Mouse G900
Keyboard Corsair k60 RGB PRO
Software Win 11
In hardocp's call of juarez test the 2900 took less of a hit with aa on than a gts, so it cant be all bad
 

kwchang007

New Member
Joined
Jan 27, 2007
Messages
3,962 (0.63/day)
Location
Severn, MD, USA.
Processor C2D T7200@2 ghz vcore: .9875
Motherboard generic laptop board
Cooling fan control and antec notebook cooler
Memory 2 GBs@ 533 mhz ddr2
Video Card(s) x1400 mobile, overclocked: 526mhz core/ 882mhz ddr
Storage 120 GB@ 5400 rpm fujitsu
Display(s) 15.4" 1440x900
Audio Device(s) integrated
Software vista 32 bit home premium
@grings...it has a larger bus size. Shouldn't it be better at AA?
 
Top