OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 638.

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Vbs on February 23, 2012, 09:27:30 AM

Btw, why check for 'result' in the first place? Isn't that a redundant check?

Using now,

Code:

#elif defined VECTORS2
          if (!W[117].x) {
              output[FOUND] = FOUND;
	      output[NFLAG & W[3].x] = W[3].x;
          }
	  if (!W[117].y) {
              output[FOUND] = FOUND;
	      output[NFLAG & W[3].y] = W[3].y;
          }

Seems to work! Grin

No, you're doing two branches on the common path now. We want as few branches as possible in the common path. It doesn't matter if we do 2 extra checks in the uncommon (i.e. found share) path.

jamesg

vip

Activity: 1358

Merit: 1000

AKA: gigavps

I have posted a bounty to get better documentation around the --scan-serial option for linux -> https://bitcointalksearch.org/topic/bounty-5-btc-clear-instructions-for-cgminers-scan-serial-option-65879

If you have this knowledge, you can make some quick btc... Wink

Vbs

hero member

Activity: 504

Merit: 500

Btw, why check for 'result' in the first place? Isn't that a redundant check?

Using now,

Code:

#elif defined VECTORS2
          if (!W[117].x) {
              output[FOUND] = FOUND;
	      output[NFLAG & W[3].x] = W[3].x;
          }
	  if (!W[117].y) {
              output[FOUND] = FOUND;
	      output[NFLAG & W[3].y] = W[3].y;
          }

Seems to work! Grin

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Vbs on February 23, 2012, 09:00:04 AM

OK, fixed it:
"bool result = any(W[117] == 0)" seems to be the slow op, changing back to the bitwise AND "W[117].x & W[117].y" (and the next if to !result), gets me back at ~400MH/s.

Goddamn. Thanks for that. The any() function is meant to be faster than manually unrolling it the way I did in the old version. I guess I shouldn't put any faith in their hardware functions...

sigh...

Vbs

hero member

Activity: 504

Merit: 500

OK, fixed it:
"bool result = any(W[117] == 0)" seems to be the slow op, changing back to the bitwise AND "W[117].x & W[117].y" (and the next if to !result), gets me back at ~400MH/s.

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Vbs on February 23, 2012, 08:50:37 AM

Ok, my findings so far:
5850 on SDK 2.5

2.2.7 ~400MH/s with phatk120213

2.3.0 ~320MH/s with phatk120222
2.3.0 ~400MH/s with phatk120213 *but* lots of HW errors!

So it seems the hashrate drop is in the kernel changes? Huh

The old kernel is simply not compatible with the current one.

Well this makes me fucking angry. All I can do is shake my fist even harder at ATI for the changes just SHOULD NOT CAUSE THIS as they're meant to be trivial changes. Well more fucking quick fucking releases to deal with fucking ATI fucking fail. Rollback phatk fucking time.

Vbs

hero member

Activity: 504

Merit: 500

Ok, my findings so far:
5850 on SDK 2.5

2.2.7 ~400MH/s with phatk120213

2.3.0 ~320MH/s with phatk120222
2.3.0 ~400MH/s with phatk120213 *but* lots of HW errors!

So it seems the hashrate drop is in the kernel changes? Huh

cablepair

hero member

Activity: 896

Merit: 1000

Buy this account on March-2019. New Owner here!!

thank you ckvolias for this big release on my Birthday!

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: stevegee58 on February 23, 2012, 08:03:39 AM

Quote from: tenzor on February 23, 2012, 08:01:48 AM

Quote from: stevegee58 on February 23, 2012, 07:57:39 AM

My 6770 runs waaaaaaaaaaaaaaaaay faster with the diablo kernel.

I was reliably getting around 185-190 Mhash/sec with poclbm. Now it's reliably around 202 with diablo. Really noice!

Hmm. my 6770 runs around 230 Nhash/sec with phatk at 1030/300 clocks. -v2 -w256

I haven't been able to reliably run my 6770 above 960.

That doesn't mean you can't try -k phatk -v 2 -w 256
Based on what tenzor says, 960/1030 * 230 means you should be able to get 214

edit: He changed that to 220-225? * 960/103 is 205 which is not far off what you're getting... Try it anyway Wink

edit2: DONT FORGET SDK MATTERS!

tenzor

sr. member

Activity: 308

Merit: 250

Quote from: stevegee58 on February 23, 2012, 08:03:39 AM

I haven't been able to reliably run my 6770 above 960.

Ok. @960/300 I have 220-225
cgminer 2.2.6

Edit:
Summary:
@960/300 = 220-225
@1000/300 = around 230

cgminer 2.2.6
-I 9 -v 2 -w 256 -k phatk
SDK 2.4, ubuntu, catalyst 11.6

stevegee58

legendary

Activity: 916

Merit: 1003

Quote from: tenzor on February 23, 2012, 08:01:48 AM

Quote from: stevegee58 on February 23, 2012, 07:57:39 AM

My 6770 runs waaaaaaaaaaaaaaaaay faster with the diablo kernel.

I was reliably getting around 185-190 Mhash/sec with poclbm. Now it's reliably around 202 with diablo. Really noice!

Hmm. my 6770 runs around 230 Nhash/sec with phatk at 1030/300 clocks. -v2 -w256

I haven't been able to reliably run my 6770 above 960.

tenzor

sr. member

Activity: 308

Merit: 250

Quote from: stevegee58 on February 23, 2012, 07:57:39 AM

My 6770 runs waaaaaaaaaaaaaaaaay faster with the diablo kernel.

I was reliably getting around 185-190 Mhash/sec with poclbm. Now it's reliably around 202 with diablo. Really noice!

Hmm. my 6770 runs around 230 Nhash/sec with phatk at 1030/300 clocks. -v2 -w256

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: stevegee58 on February 23, 2012, 07:57:39 AM

My 6770 runs waaaaaaaaaaaaaaaaay faster with the diablo kernel.

I was reliably getting around 185-190 Mhash/sec with poclbm. Now it's reliably around 202 with diablo. Really noice!

Excellent, that's more the sort of news I was hoping for Grin

stevegee58

legendary

Activity: 916

Merit: 1003

My 6770 runs waaaaaaaaaaaaaaaaay faster with the diablo kernel.

I was reliably getting around 185-190 Mhash/sec with poclbm and phatk. Now it's reliably around 202 with diablo. Really noice!

I did notice that suddenly my poclbm performance dropped from around 185 to like 165 with this new release. But I'm not complaining because of the pleasant diablo surprise.

Vbs

hero member

Activity: 504

Merit: 500

Quote from: -ck on February 23, 2012, 07:36:18 AM

Quote from: Vbs on February 23, 2012, 07:32:39 AM

I now have one 5850 running cgminer 2.2.7 and another on 2.3.0, on the same engine/mem clocks. I'll be counting the accepted shares on the pool they are connected to and see if there is a significant difference. Cgminer 2.2.7 is reporting ~400MH/s and 2.3.0 is reporting ~320MH/s atm. At first glance, it seems they are submitting the same amount of shares (39 vs 40), so this could just be a "cosmetic" change in cgminer 2.3.0. Grin

That, unfortunately, does not make sense... and this hashrate drop is more than a little disturbing since there is no real valid explanation for it.

edit: it's so big it's like one has an intensity set and the other is running dynamic.

Yeah, you're right, it was just wishful thinking... The one on 2.3.0 has a U:4.22 vs 2.2.7 with U:5.35. Both are running at I:9.

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Vbs on February 23, 2012, 07:32:39 AM

I now have one 5850 running cgminer 2.2.7 and another on 2.3.0, on the same engine/mem clocks. I'll be counting the accepted shares on the pool they are connected to and see if there is a significant difference. Cgminer 2.2.7 is reporting ~400MH/s and 2.3.0 is reporting ~320MH/s atm. At first glance, it seems they are submitting the same amount of shares (39 vs 40), so this could just be a "cosmetic" change in cgminer 2.3.0. Grin

That, unfortunately, does not make sense... and this hashrate drop is more than a little disturbing since there is no real valid explanation for it.

edit: it's so big it's like one has an intensity set and the other is running dynamic.

Vbs

hero member

Activity: 504

Merit: 500

I now have one 5850 running cgminer 2.2.7 and another on 2.3.0, on the same engine/mem clocks. I'll be counting the accepted shares on the pool they are connected to and see if there is a significant difference. Cgminer 2.2.7 is reporting ~400MH/s and 2.3.0 is reporting ~320MH/s atm. At first glance, it seems they are submitting the same amount of shares (39 vs 40), so this could just be a "cosmetic" change in cgminer 2.3.0. Grin

Vbs

hero member

Activity: 504

Merit: 500

Quote from: -ck on February 23, 2012, 07:10:19 AM

Quote from: Vbs on February 23, 2012, 07:07:19 AM

The cgminer window closes really fast on Windows, you really need to put an #ifdef win getch() at the end of execution, since piping the output doesn't work either Tongue

Anyway, by a screen grab, it shows Platform version: OpenCL 1.1 AMD-APP-SDK-v2.5 (793.1), with 1 platform device (GPU 0 ATI Radeon HD 5800 Series), and my settings for engine (970) and mem (150).

The default kernel is phatk120222Cypressv2w128l4.

Now that's a serious groaner since the phatk kernel is only trivially changed it should not perform any differently Huh

Try -v 2 -w 256 since you're grossly underclocking memory.

If you start it from a dos prompt window the window won't close...

V2w256 is what I always use to get the ~400MH/s, but it seems to hover around ~320MH/s now. Gonna let it run more time and see if the stats change.

I did start it from a dos prompt, but it creates a new (console) window and closes it after it's done. Piping the output like "cgminer -n > report.txt" doesn't work either. Windows really sucks on these small things :S

-ck

legendary

Activity: 4088

Merit: 1631

Ruu \o/

Quote from: Vbs on February 23, 2012, 07:07:19 AM

The cgminer window closes really fast on Windows, you really need to put an #ifdef win getch() at the end of execution, since piping the output doesn't work either Tongue

Anyway, by a screen grab, it shows Platform version: OpenCL 1.1 AMD-APP-SDK-v2.5 (793.1), with 1 platform device (GPU 0 ATI Radeon HD 5800 Series), and my settings for engine (970) and mem (150).

The default kernel is phatk120222Cypressv2w128l4.

Now that's a serious groaner since the phatk kernel is only trivially changed it should not perform any differently Huh

Try -v 2 -w 256 since you're grossly underclocking memory.

If you start it from a dos prompt window the window won't close...

Vbs

hero member

Activity: 504

Merit: 500

The cgminer window closes really fast on Windows, you really need to put an #ifdef win getch() at the end of execution, since piping the output doesn't work either Tongue

Anyway, by a screen grab, it shows Platform version: OpenCL 1.1 AMD-APP-SDK-v2.5 (793.1), with 1 platform device (GPU 0 ATI Radeon HD 5800 Series), and my settings for engine (970) and mem (150).

The default kernel is phatk120222Cypressv2w128l4.

Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 638. (Read 5805546 times)