Topic: OFFICIAL CGMINER mining software thread for linux/win/osx/mips/arm/r-pi 4.11.0 - page 633. (Read 5805975 times)

Believe it or not I can actually shed some light on this, and since I'm in a much better mood as someone just kindly donated some BTC, I'll even answer in less than my usual appalling AMD-induced tone of late.

One of the things about the 69xx and now the 79xx architectures is the ability internally to underclock memory relative to running speed. There is a power-usage war going on now between manufacturers, and this is one place where AMD is working very hard (unlike fixing drivers, SDKs and so on). Since the GPUs are running the ram double channel, if the RAM bandwidth is not in full use, they can shut down one of the channels. This is why the power usage does not appear to be directly proportional to the RAM speed. So even though you can only decrease your clock speed to say 900, it might be internally running at 450. This is also why when you flash the bios and underclock the ram, it might crash at apparently satisfactory rates and 6970s are virtually never stable below 300 whereas 5xxx can happily run down to 150. Sure there is more power to be saved if you can actually flash their bios and turn them down to 300 since you are guaranteed to never actually jump between 300 and 600, but it is not universally half the ram speed and power consumption.  Bear in mind that most people do not touch the clock speed of their memory (except usually to increase it) but they do care about power consumption. This is also why it's so hard to pin down power usage on these things as they fluctuate wildly depending on the type of load rather than just the overall load. 100% GPU load could really mean anything and might or might not be high ram bandwidth.

http://sites.amd.com/de/Documents/PowerTune_Technology_Whitepaper.pdf

It's nothing you should change as long as everything works ... in border case situations a + to the PowerTune setting would lead to possible higher power consumption and perhaps a tad more performance, but I think AMDs defaults are fine.

Dia

This.

There are two ways to control a videocard:
* the right way
* and the hack way

The "right" way:
Using the AMD driver library you can send requests tot he card.  You can't control anything.  The card is free to ignore or modify any request as it sees fit.  

So it is more like this:
cgminer (via driver): "Video card #1 can you please change clock to 300 Mhz".
video card #1: "command is valid"
internal BIO check. 300Mhz is invalid, ignoring.

So why does AB "show" 300Mhz.  It doesn't.

AB, GPU-Z (main tab), Trixx, etc show what the card has been SET TO not what it is RUNNING AT.

The only three places I have found to always report correct values on what the card IS ACTUALLY RUNNING AT are:
* cgminer.  If cgminer says a card is running at 300Mhz, 30 Ghz, or 0.1 Mhz it probably is right.  It doesn't matter how weird you may think that is.  I can't remember a single instance where cgminer turned out to be wrong.
* (windows) GPU-Z Sensor tab.  Note the first tab shows what card is set at.  That is useless.  If you look on sensor tab it shows what card is running at.
* (linux) aticonfig

The "hack" way:
So how do tools like AB change the clock?  They bypass the drivers and write directly to the GPU BIOS.  This is why it often requires a new version before AB will work with newly released cards.  This also applies to things like how GPU-Z can read VRM temps or how radeonvolt can modify voltage beyond what is allowed by drivers.  The problem with this method is that it isn't universal.  RadeonVolt doesn't work on 5970 so I can't modify my voltage above what is allowed by GPU BIOS.  Maybe someday someone will hack a solution together for Linux, maybe they never will.  

The annoyance:
Since so many hacks exist but aren't universal and bypass the drivers one would think AMD would expand the official drivers to allow full range of clock adjustments, voltage adjustments, and sensor data readings.  Of course they won't.  In the meantime you can flash a card with custom bios to make it do just about anything.  Run at 100Mhz memclock, have "stock" core speed of 1.2Ghz, have a core voltage of 0.7V, etc.  Granted you can also completely destroy the card in an non-warrantied manner but it is possible to change just about anything.

cgminer is not doing any adjustment of anything. It sends the request to the driver. The driver says it has accepted the value for the profile. The hardware then gladly ignores you and although the profile now says the memory is 300, the GPU goes back to its default speed. This is why I made cgminer report back the actual values to you after you try to make a change. If it doesnt work it doesnt work. Nothing can make cgminer make it work because it doesnt have access to the special hardware backdoor commands that afterburner and co. can fuck the operating system up the arse with. AMD did not release a public library for anal reaming of GPUs.

Afterburner doesn't work anymore if using the latest (12.3) drivers from AMD.
I'm not sure if it's still working with older versions of catalyst, since I just installed my 7970, and the latest drivers are all I've used.

You could try to use MSI Afterburner, see here: http://forums.guru3d.com/showthread.php?t=358990 and enable it's unofficial overclocking mode, see here: http://forums.guru3d.com/showthread.php?t=338906

Dia

No. cgminer uses the ATI Display Library to set clock speeds. You need specific windows-only tools to clock memory lower than the engine -150 limit. I don't have access to those functions to use with cgminer.

I just want to reiterate my results with the diablo kernel under cgminer 2.3.1-2, SDK 2.6, WinXP.
I am able to consistently get around 207 mhs with diablo at dynamic intensities of 4 to 5 and -v 2 -w 256.
In order to obtain the same hash rate with phatk I had to run a fixed intensity of 9 with -v 2 -w 256.
I can no longer coax poclbm above about 165.  Before 2.3.x, poclbm was my best performer at around 185 mhs with no special settings.

Based on some reading in the diablo thread, it looks like it's actually optimized for the 2.6 SDK.  This probably explains why it works so well for me.

Latest package should be used as Con reverted negative kernel changes. If further anomalies arise post here again .

Dia

So which one should the typical miner be using, and does each one have separate kernel versions now?  Also, for the 2.3.0-1 kernels I e-mailed you, should they be ID'd as such somehow?

Yes, yes, yes, yes. When I'm using 2.3.0 all works perfectly!

Yay, I have my hashes back now

Thanks for using, I'm responsible for most changes there ... did you specify -v 2 and -w 256 as this is needed for best performance with diakgcn? Did you delete the .bin files? Every change I make is checked via KernelAnalyzer and a look at the GPU ISA code. Have you the latest OpenCL runtime installed, as that's needed, too.

Dia

It's in fact back to normal (well, 2.2.7 speeds anyway) now that it's been hashing for a few minutes and levelled off.

As a side note, out of pure desperation, I tried the Diablo kernel when testing 2.3.1 original and it was very close in speeds/performance to the 2.2.7 packaged phatk (with my 6950's, CAT 12.1, SDK 2.4......it wouldn't start though with 5830's, CAT 11.8, SDK 2.1)....lol

thanks again,
bitlane

Well.. it's exactly the same kernel as 2.2.7 so it MUST be the same performance Anything else is just reporting hashrate differences. Thanks for testing.

I've upgrade the package to a 2.3.1-2 package, doing the same change to ALL the kernels in case someone else is affected.

Enjoy.

HUGE WIN !

5830, SDK 2.1, CAT 12.1...... only 5 MH/s slower per card than 2.2.7 (was 10+ MH/s slower using 2.3.0 & 2.3.1)

6950, SDK 2.4, CAT 12.1...... only 10 MH/s slower per card than 2.2.7 (was 70 MH/s slower using 2.3.0 & 2.3.1)

This is a huge help,
thanks

Stop using diakgcn with 7970. -k poclbm is fastest.

Performance down for 7970 with diakgcn from 550 to 410 

No it's just a name. If this is true then the thing causing the regression is the most unlikely thing in the universe and AMD is going into another dimension with its fail. I'm going to beat this mother fucking piece of shit if it's the last thing I ever do. Redownload the cgminer-2.3.1-1 version please.

Win7 x64
Driver = 12.1 Cat, 2.1 SDK
5x HD5830 Cards

I start CGMiner with a BAT file and only add CLOCK, NETWORK and INTENSITY ( 8 ) settings. Everything else is default (no kernel specified, work size etc....nothing).
................................................
2.2.7 generates bin = phatk120213Cypressbitalignv2w128long4.BIN
2.3.1 generates bin = phatk120223Cypressv2w128l4.BIN
................................................

Using 2.3.1, Performance has been negatively affected compared to 2.2.7 (the same can be seen using 2.3.0-1). The most dramatic performance hit can be seen with my 6950's, by 70+MH/s per card loss of performance, same as was with 2.3.0-1 earlier.
Does the omission of 'bitalign' in the BIN file name have something to do with this ?

bitlane.