GeForce 9800 GTX/9800 GTX+ (driver version 19562, CUDA version 3000, compute capability 1.1, 512MB, 484 GFLOPS peak)

---

GeForce 9500 GT (driver version unknown, CUDA version 2030, compute capability 1.1, 512MB, 86 GFLOPS peak)

This is reporting a much faster peak rate than previous entries for this card.

Running from Linux. PCI-E x16 bus.

GeForce 9500 GT (driver version unknown, CUDA version 2030, compute capability 1.1, 512MB, 86 GFLOPS peak)

This is reporting a much faster peak rate than previous entries for this card.

Running from Linux. PCI-E x16 bus.

Actually, both Wonder Puppy and perryjay have reported higher peak figures (106 and 118 GFLOPS respectively - the larger memory shouldn't affect the speed)

The lower figures have all been 'est' GFLOPS.

Richard, actually mine is OCd a bit. At stock speed it reports 90 GFLOPS peak. I have 1024MB memory. It is OCd to ~1850shader/ 725core. Memory is still at 400.

---

PROUD MEMBER OF Team Starfire World BOINC

GeForce 9500 GT (driver version unknown, CUDA version 2030, compute capability 1.1, 512MB, 86 GFLOPS peak)

This is reporting a much faster peak rate than previous entries for this card.

Running from Linux. PCI-E x16 bus.

Actually, both Wonder Puppy and perryjay have reported higher peak figures (106 and 118 GFLOPS respectively - the larger memory shouldn't affect the speed)

The lower figures have all been 'est' GFLOPS.

My 9500GT is slightly oc'd: core 753 / shader 1774 / memory 595
2009-12-12 12:18:13 PM NVIDIA GPU 0: GeForce 9500 GT (driver version 19562, CUDA version 3000, compute capability 1.1, 1024MB, 114 GFLOPS peak)

temps while crunching run 70c-74c

---

OK guys! Repeat after me "My name is ... and I am a BOINCaholic" ;-)

All I was doing is pointing out that we have (still) to be careful about the difference between 'est. GFLOPS' and 'GFLOPS peak', and never try to compare the two without that 5.6x correction factor.

24.12.2009 10:20:29 NVIDIA GPU 0: GeForce 9600 GT (driver version 19581, CUDA version 3000, compute capability 1.1, 512MB, 208 GFLOPS peak)

---

12/26/2009 NVIDIA GPU 0: GeForce 8400 GS (driver version 19562, CUDA version 3000, compute capability 1.1, 256MB, 29 GFLOPS peak)

---

23-12-2009 18:50:05 Data directory: C:\Program Files\BOINC
23-12-2009 18:50:05 Running under account Administrator
23-12-2009 18:50:05 Processor: 4 GenuineIntel Intel(R) Core(TM)2 Extreme CPU X9650 @ 3.00GHz [EM64T Family 6 Model 23 Stepping 6]
23-12-2009 18:50:05 Processor: 6.00 MB cache
23-12-2009 18:50:05 Processor features: fpu tsc pae nx sse sse2
23-12-2009 18:50:05 OS: Microsoft Windows XP: Professional x64 Edition, Service Pack 2, (05.02.3790.00)
23-12-2009 18:50:05 Memory: 4.00 GB physical, 5.75 GB virtual
23-12-2009 18:50:05 Disk: 232.88 GB total, 193.84 GB free
23-12-2009 18:50:05 Local time is UTC +1 hours
23-12-2009 18:50:05 NVIDIA GPU 0: GeForce 9800 GTX/9800 GTX+ (driver version 19562,
CUDA version 3000, compute capability 1.1, 512MB, 484 GFLOPS peak)Was 85GFLOPS, measured with BOINC 6.4.5 or lower, since 6.10.xx it looks like a different measurement!
23-12-2009 18:50:05 SETI@home Found app_info.xml; using anonymous platform.
PCI-E x16 1.0; slot however is capable of PCI-E(x16) 2.0

And another host:
23-12-2009 9:15:38 Starting BOINC client version 6.10.15 for windows_intelx86
23-12-2009 9:15:38 log flags: file_xfer, sched_ops, task
23-12-2009 9:15:38 Libraries: libcurl/7.19.4 OpenSSL/0.9.8k zlib/1.2.3
23-12-2009 9:15:38 Data directory: C:\Documents and Settings\All Users\Application Data\BOINC
23-12-2009 9:15:38 Running under account Gebruiker
23-12-2009 9:15:38 Processor: 4 GenuineIntel Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz [x86 Family 6 Model 15 Stepping 11]
23-12-2009 9:15:38 Processor: 4.00 MB cache
23-12-2009 9:15:38 Processor features: fpu tsc pae sse sse2 mmx
23-12-2009 9:15:38 OS: Microsoft Windows XP: Professional x86 Edition, Service Pack 3, (05.01.2600.00)
23-12-2009 9:15:38 Memory: 2.00 GB physical, 3.85 GB virtual
23-12-2009 9:15:38 Disk: 195.31 GB total, 176.64 GB free
23-12-2009 9:15:38 Local time is UTC +1 hours
23-12-2009 9:15:38 ATI GPU 0: ATI Radeon HD5700 series (Juniper) (CAL version 1.4.427, 1024MB, 1360 GFLOPS peak)
23-12-2009 9:15:38 SETI@home Found app_info.xml; using anonymous platform.
ATI 5770; PCI-E(x16) 2.0.

---

12/28/2009 7:47:14 AM Processor: 8 GenuineIntel Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz [Intel64 Family 6 Model 26 Stepping 5]
12/28/2009 7:47:14 AM Processor: 256.00 KB cache
12/28/2009 7:47:14 AM Processor features: fpu tsc pae nx sse sse2 pni mmx
12/28/2009 7:47:14 AM OS: Microsoft Windows 7: Home Premium x64 Edition, (06.01.7600.00)
12/28/2009 7:47:14 AM Memory: 5.99 GB physical, 11.98 GB virtual
12/28/2009 7:47:14 AM Disk: 931.50 GB total, 850.87 GB free
12/28/2009 7:47:14 AM Local time is UTC -5 hours
12/28/2009 7:47:18 AM NVIDIA GPU 0: GeForce GTS 250 (driver version 19562, CUDA version 3000, compute capability 1.1, 1024MB, 456 GFLOPS peak)

29/12/2009 08:00:50 NVIDIA GPU 0: GeForce 8800 GT (driver version 19107, CUDA version 2030, compute capability 1.1, 512MB, 351 GFLOPS peak)

---

NVIDIA GPU 0: GeForce GTX 260 (driver version 19562, CUDA version 3000, compute capability 1.3, 896MB, 537 GFLOPS peak)

1/1/2010 8:42:36 PM NVIDIA GPU 0: GeForce 8800 GTS (driver version 19562, CUDA version 3000, compute capability 1.0, 320MB, 228 GFLOPS peak)
1/1/2010 8:42:36 PM NVIDIA GPU 1: GeForce 8600 GTS (driver version 19562, CUDA version 3000, compute capability 1.1, 256MB, 93 GFLOPS peak)

I have a GeForce 8400 PCI that seems to get the same 22 GFLOPS peak as my GeForoce 8400 in PCIeX16 . . . In different PCs...

The PCI 8400 is running with a 9600GSO card in PCIeX16. BOINC seems to think the 9600GSO can do 264 GFLOPS peak with driver 19562.

Another PC with the same 9600GSO Card (ASUS) the BOINC client gets 47 GFLOPS with the 19107 & 19562 driver.

?????
*sigh*

---

1/1/2010 8:42:36 PM NVIDIA GPU 0: GeForce 8800 GTS (driver version 19562, CUDA version 3000, compute capability 1.0, 320MB, 228 GFLOPS peak)
1/1/2010 8:42:36 PM NVIDIA GPU 1: GeForce 8600 GTS (driver version 19562, CUDA version 3000, compute capability 1.1, 256MB, 93 GFLOPS peak)

I have a GeForce 8400 PCI that seems to get the same 22 GFLOPS peak as my GeForoce 8400 in PCIeX16 . . . In different PCs...

The PCI 8400 is running with a 9600GSO card in PCIeX16. BOINC seems to think the 9600GSO can do 264 GFLOPS peak with driver 19562.

Another PC with the same 9600GSO Card (ASUS) the BOINC client gets 47 GFLOPS with the 19107 & 19562 driver.

?????
*sigh*

There are only two characteristics of CUDA cards considered in the flops calculations; "clockRate" and "multiProcessorCount". Those are multiplied, then new versions of BOINC scale by 16000 to give the peak value. Older versions of BOINC scaled by about 2857 to give an estimate based on comparative measurements of an FX3700 vs. some unknown CPU.

                                                            Joe

Konata Izumi wrote:

BOINC seems to think the 9600GSO can do 264 GFLOPS peak with driver 19562.

Another PC with the same 9600GSO Card (ASUS) the BOINC client gets 47 GFLOPS with the 19107 & 19562 driver.

?????
*sigh*

Boinc 6.10.x estimates the Flops differant (higher) than Boinc 6.6.x
I see one of your computers with the 9600GSO has 6.6.x and the rest have 6.10.x

---

Konata Izumi wrote:

BOINC seems to think the 9600GSO can do 264 GFLOPS peak with driver 19562.

Another PC with the same 9600GSO Card (ASUS) the BOINC client gets 47 GFLOPS with the 19107 & 19562 driver.

?????
*sigh*

Boinc 6.10.x estimates the Flops differant (higher) than Boinc 6.6.x
I see one of your computers with the 9600GSO has 6.6.x and the rest have 6.10.x

But remember that the version of BOINC in use makes no difference at all to the actual processing speed: just reporting some numbers using a different scale doesn't change the underlying reality.