After a series of driver crashes, I installed cat.11.4 (+AMD-SPP-SDK 2.4), I also
did not have OpenCL support, except the I7

Tuning the new installer is necessary too. I've set Iterations for pulsefind to
2, 1 gave a heavy screen lag and run 2 Instances_per-device
hope to get enough MB work.

AstroPulse, only changed unroll_data_chunk to 15, 1 per device and
default ffa_block and ffa_block_fetch, 4096 and 2048.

The GPU load on MB work is ~50%, also heavely fluctuating, 1% - 96%,
I've ran every BenchMark Test with SiSoftSandra no problems.

Ehh, another driver-crash, I'll add the ' BSOD' messages, in a moment, as
I'm typing this from my LT. (Also running BOINC -SETI)

Problem signature:
  Event :	BlueScreen
  Windows version:	6.1.7601.2.1.0.256.48
  Country-id:	1043

Extra information on this event:
  BCCode:	116
  BCP1:	FFFFFA800D1534E0
  BCP2:	FFFFF880016E2740
  BCP3:	0000000000000000
  BCP4:	0000000000000002
  OS Version:	6_1_7601
  Service Pack:	1_0
  Product:	256_1

Files which can help in solving the problem:
  C:\Windows\Minidump\120311-43539-01.dmp
  C:\Users\FredJVerster\AppData\Local\Temp\WER-73008-0.sysdata.xml

Lees de onlineprivacyverklaring:
  http://go.microsoft.com/fwlink/?linkid=104288&clcid=0x0413

If the onlineprivacystatement isn't available, read our  offlineprivacyverklaring:
  C:\Windows\system32\nl-NL\erofflps.txt 

I've also changed my CPU capabillities to SSE3, not SSSE3x, which
my 2 other (NVIDIA 470&480) hosts do run.

Has anyone found a better driver then 11.4, as with 11.11, I've no
OpenCL support with 11.11?

---

Has anyone found a better driver then 11.4, as with 11.11, I've no
OpenCL support with 11.11?

I use Cat 11.9 and i find it ok.

I run Win Vista 32bit and have an ATI 5850

---

TRuEQ & TuVaLu

If its of any help I run Win 7 Ultimate 64, and use Cat 11.10 on my 5850. GPUZ has the Open CL & Direct Compute 5 boxes ticked.

The following drivers are known to be incompatible with the app:
11.10 and 11.11.

---

With each crime and every kindness we birth our future.

I tried 41g and I went back to 38g as I feel 38 performs better than 41.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

I tried 41g and I went back to 38g as I feel 38 performs better than 41.

You're welcome. At least x38g doesn't have the triplet-finding bug which triggered the updated release in the first place.

For other readers: please don't go back to the other, older, MultiBeam applications in the previous v0.38 installer - there's a risk of inaccurate results.

I tried 41g and I went back to 38g as I feel 38 performs better than 41.

and since your GTX 295 is computeCap 1.3, it won't suffer from the FMAD Bug eithier, (that x41g fixes for GPUs with <cc1.3)

[Update: x41f] replaced nVidia's shortcut for the <cc1.3 chirp with full df64, to reduce inconclusives. NO FMAD Allowed!!!

Claggy

I tried 41g and I went back to 38g as I feel 38 performs better than 41.

You're welcome. At least x38g doesn't have the triplet-finding bug which triggered the updated release in the first place.

For other readers: please don't go back to the other, older, MultiBeam applications in the previous v0.38 installer - there's a risk of inaccurate results.

If I could I'd replace 41G with 38G in the package and yes I tried, 7zip can read what's in there, but not alter anything. And yes on the apps, I'd say don't go back.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

I tried 41g and I went back to 38g as I feel 38 performs better than 41.

and since your GTX 295 is computeCap 1.3, it won't suffer from the FMAD Bug eithier, (that x41g fixes for GPUs with <cc1.3)

Claggy

Ok, So My 295 card won't go fMAD then. ;)

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

it'a just a matter of copy/pasting the section of app_info and apps that you want to use for the older app while using the newer ones on your cpu. just remember to remove the app_info portions that point to the new GPU app and you should be fine

---

In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope

I noticed the P3 application AK_v8b_win_SSE_p3.exe didn't get the b2 treatment. Is it just a typo in the file name or is the exe unchanged for P3s?

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

I noticed the P3 application AK_v8b_win_SSE_p3.exe didn't get the b2 treatment. Is it just a typo in the file name or is the exe unchanged for P3s?

That exe is unchanged for P3s. Quoting Joe Segur, whose sharp eyes spotted an inconclusive result and who had the tenacity to follow it up:

I just remembered that AK_v8b_win_SSE_p3.exe doesn't have many of the v8b changes, so is effectively an AK_v8 build (the full AK_v8b build was slower). A quick bench test confirms that it gets "Result : Strongly similar, Q= 99.30%" against the 6.03 reference.

I noticed the P3 application AK_v8b_win_SSE_p3.exe didn't get the b2 treatment. Is it just a typo in the file name or is the exe unchanged for P3s?

The SSE_p3 app didn't have the Bug, as it is from before the Bug was introduced.

Claggy

Running x39e with 280.26

Intend to upgrade over weekend, First bit of free time.

One thing that I have not seen mention so far is downclocking, how is this release with regards to downclocking with the last few releases of drivers?

This is fixed in the new release.

Upgraded, new app_info files done, all running good.

I keep an app_info file for 1, 2 and 3 files per card and just copy and paste in as required.

---

Kevin

I have run out of GPU WU's so am trying out VLAR's on x41g.

Processing 1 per card speed looks about the same or slightly better than x39e, no serious slow down on machine.

If I get any problems I will let you know.

---

Kevin

So, Q8200: NO, NO, NO, ION: inconclusive, and i3: Haven't got a clue yet.

Thanks for devoting time and energy to the research. I guess nothing so far is too surprising: the Core2 range, like your Q8200, were exactly the target platform for the SSSE3x builds, so it's good to have confirmed that nothing has broken there.

The really interesting one will be the i3, once the servers have condescended to co-operate.

I just installed the new Lunatic apps and when I got my first cuda WU it ran for about 2 minutes without showing any progress and then my driver crashed. Any idea how to fix that problem?

GPU: Geforce G210 (512MB)
Driver: 280.26
OS: Windows7 32bit

---

I just installed the new Lunatic apps and when I got my first cuda WU it ran for about 2 minutes without showing any progress and then my driver crashed. Any idea how to fix that problem?

GPU: Geforce G210 (512MB)
Driver: 280.26
OS: Windows7 32bit

Could you report the crashed task, please, so that we can inspect stderr_txt?

I just installed the new Lunatic apps and when I got my first cuda WU it ran for about 2 minutes without showing any progress and then my driver crashed. Any idea how to fix that problem?

GPU: Geforce G210 (512MB)
Driver: 280.26
OS: Windows7 32bit

Could you report the crashed task, please, so that we can inspect stderr_txt?

It didn't crash, it just stopped. I now tried to resume, it started over from 00:00 and now seems to run fine (22% done atm). I let this one finish now and see how the next one works.

---

I have run out of GPU WU's so am trying out VLAR's on x41g.

Processing 1 per card speed looks about the same or slightly better than x39e, no serious slow down on machine.

If I get any problems I will let you know.

Finaly finished uploading, now to fill cache up again.

Processed 20 VLAR's on GPU's, all now uploaded and verified, average time per WU 64 min, no problems with processing seen.

I had slight stuttering when playing avi files, but that is to be expected, other than that no noticeable affects.

---

Kevin

So, Q8200: NO, NO, NO, ION: inconclusive, and i3: Haven't got a clue yet.

Thanks for devoting time and energy to the research. I guess nothing so far is too surprising: the Core2 range, like your Q8200, were exactly the target platform for the SSSE3x builds, so it's good to have confirmed that nothing has broken there.

The really interesting one will be the i3, once the servers have condescended to co-operate.

I now have a VLAR on the i3 that I will run against a similar VLAR on the Q8200 in a few minutes. That can be interesting, lets just hope that they both run full time.

Q8200 SSSE3x: 20oc11ab.17878.306819.10.10.125.vlar_2

i3 SSE3: task 20oc11ab.17878.307228.10.10.94.vlar_0

The q8200 takes about 3 hours + for a VLAR with SSSE3x. What will the i3 do, with SSE3, that can be an interesting race :-)

And finished.

Well, the Q8200 is beaten to the ground on this comparison anyhow. The small i3 Laptop wins. Also worth knowing, is that the Q8200 hates VHARS (yes VHARS, not VLARS) running one VHAR on the CPU is ok, but if it runs more than one at a time, they all slow down to a crawl, and the time to finishing will almost double.
The i3 does not seem to suffer from that though, it gladly chews up all the VHARs I can throw at it. Also worth knowing, is that the i3 seems to beat the Q8200 on all MB AR’s.
The Q8200 is only faster than the i3, when running AP’s, and then not much faster. The only thing that keeps the Q8200 ahead of the i3 now, is the ATI 4850 card, which of course is much faster than the NVIDIA GeForce 315M card in the i3.

Q8200 SSSE3x: 20oc11ab.17878.306819.10.10.125.vlar_2
Workunit: 2188717797
WU true angle range is : 0.011215
Run Time: 11,400.55 Seconds

i3 SSE3: task 20oc11ab.17878.307228.10.10.94.vlar_0.
Workunit: 2188468270
WU true angle range is : 0.011534
Run Time: 9,145.31 Seconds

Well, this was a little side step from what I was really going to do, comparing SSSE3X with SSE3 on the i3. That’s still on the to-do list.

I noticed that my i3-390 notebook and i7-860 HTPC handle VLAR tasks better with the new build. Where before I would see some laggyness in the system when all cores were doing VLAR tasks. It might also just been the current batch of work.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[