Bitcoin Forum
May 04, 2024, 04:36:25 AM *
News: Latest Bitcoin Core release: 27.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 [34] 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 »
  Print  
Author Topic: Custom RAM Timings for GPU's with GDDR5 - DOWNLOAD LINKS - UPDATED  (Read 155459 times)
doktor83
Hero Member
*****
Offline Offline

Activity: 2520
Merit: 626


View Profile WWW
April 23, 2017, 01:10:44 PM
 #661

Puzzled with this Rx480 Asus Dual 8Gb, same memory straps that came with my reference ones, did the same mod to the straps ( 1750 to 2000 ) and it keeps crashing.

Anyone think it could be a duff card?

Ii bet your mem is micron ? Smiley

SRBMiner-MULTI thread - HERE
http://www.srbminer.com
1714797385
Hero Member
*
Offline Offline

Posts: 1714797385

View Profile Personal Message (Offline)

Ignore
1714797385
Reply with quote  #2

1714797385
Report to moderator
Once a transaction has 6 confirmations, it is extremely unlikely that an attacker without at least 50% of the network's computation power would be able to reverse it.
Advertised sites are not endorsed by the Bitcoin Forum. They may be unsafe, untrustworthy, or illegal in your jurisdiction.
Oakey22
Sr. Member
****
Offline Offline

Activity: 488
Merit: 322


View Profile
April 23, 2017, 01:13:38 PM
 #662

Puzzled with this Rx480 Asus Dual 8Gb, same memory straps that came with my reference ones, did the same mod to the straps ( 1750 to 2000 ) and it keeps crashing.

Anyone think it could be a duff card?

Ii bet your mem is micron ? Smiley

Samsung memory Sad
laik2
Sr. Member
****
Offline Offline

Activity: 652
Merit: 266



View Profile WWW
April 23, 2017, 01:29:49 PM
 #663

Puzzled with this Rx480 Asus Dual 8Gb, same memory straps that came with my reference ones, did the same mod to the straps ( 1750 to 2000 ) and it keeps crashing.

Anyone think it could be a duff card?

Ii bet your mem is micron ? Smiley

Samsung memory Sad
Its actually samsung. I can advise you to lower your VDDCI to 850 and try again. I had one of those gpus that crashed on stock, lowered vddci and it was more stable.

Miners Mining Platform [ MMP OS ] - https://app.mmpos.eu/
doktor83
Hero Member
*****
Offline Offline

Activity: 2520
Merit: 626


View Profile WWW
April 23, 2017, 01:38:43 PM
 #664

Puzzled with this Rx480 Asus Dual 8Gb, same memory straps that came with my reference ones, did the same mod to the straps ( 1750 to 2000 ) and it keeps crashing.

Anyone think it could be a duff card?

Ii bet your mem is micron ? Smiley

Samsung memory Sad
Its actually samsung. I can advise you to lower your VDDCI to 850 and try again. I had one of those gpus that crashed on stock, lowered vddci and it was more stable.

I had the same thing on xfx cards, always crash on stock voltages, but when lowered, works ok  Roll Eyes

SRBMiner-MULTI thread - HERE
http://www.srbminer.com
nerdralph
Sr. Member
****
Offline Offline

Activity: 588
Merit: 251


View Profile
April 23, 2017, 02:09:43 PM
 #665

I see that uses "aticonfig" , I thought that stopped working with the latest drivers?

I tried going the linux route about 2-3 weeks ago and the only thing I found to change clock speed was "radeonjet" , and it maxed out at 2100 Mhz

I'm sticking to windows for now but thank you for sharing your script Smiley

For amdgpu I use ohgodatool and rocm-SMI
http://nerdralph.blogspot.ca/2017/03/amdgpu-pro-1660-on-ubuntu-kernel-4105.html

When you're ready to come back to the deep end of the pool we'll still be here. :-)
Oakey22
Sr. Member
****
Offline Offline

Activity: 488
Merit: 322


View Profile
April 23, 2017, 09:23:21 PM
 #666

Puzzled with this Rx480 Asus Dual 8Gb, same memory straps that came with my reference ones, did the same mod to the straps ( 1750 to 2000 ) and it keeps crashing.

Anyone think it could be a duff card?

Ii bet your mem is micron ? Smiley

Samsung memory Sad
Its actually samsung. I can advise you to lower your VDDCI to 850 and try again. I had one of those gpus that crashed on stock, lowered vddci and it was more stable.

Hi, I have now put 2 offsets in the bios, VDDC and VDDCI that wasn't there before, i have set VDDC to F8 and VDDCI to F3, which gives me a lot more stability on the system over all but 1 of the cards. I do still have 1 card on the rig that is playing up though and keeps crashing on anything other than the stock 2000 clocks. I will keep trying different settings but not holding out much hope for this one now.
PcChip
Sr. Member
****
Offline Offline

Activity: 418
Merit: 250


View Profile
April 24, 2017, 12:42:41 PM
 #667

I see that uses "aticonfig" , I thought that stopped working with the latest drivers?

I tried going the linux route about 2-3 weeks ago and the only thing I found to change clock speed was "radeonjet" , and it maxed out at 2100 Mhz

I'm sticking to windows for now but thank you for sharing your script Smiley

For amdgpu I use ohgodatool and rocm-SMI
http://nerdralph.blogspot.ca/2017/03/amdgpu-pro-1660-on-ubuntu-kernel-4105.html

When you're ready to come back to the deep end of the pool we'll still be here. :-)


on the rx470 in question I ended up hard-coding mem/core clocks & voltages in the BIOS so no tool is needed to change speed, happy with it on Win10 LTSB for now but may try linux again, thanks Smiley

question: on your blog, in the amdgpu screenshot it shows that 2100 MHz is the max, why is overclocking limited in linux? (or do you just not care becuause 2100MHz happens to be perfect for eth using your tightened timings?)

Legacy signature from 2011: 
All rates with Phoenix 1.50 / PhatK
5850 - 400 MH/s  |  5850 - 355 MH/s | 5830 - 310 MH/s  |  GTX570 - 115 MH/s | 5770 - 210 MH/s | 5770 - 200 MH/s
Q_R_V
Sr. Member
****
Offline Offline

Activity: 428
Merit: 250


Inactivity: 8963


View Profile
April 24, 2017, 01:33:59 PM
 #668

[...]
question: on your blog, in the amdgpu screenshot it shows that 2100 MHz is the max, why is overclocking limited in linux? [...]

It's not limited in linux, but it's limited in cards bios. You need to increase max mem clock in there.
nerdralph
Sr. Member
****
Offline Offline

Activity: 588
Merit: 251


View Profile
April 24, 2017, 02:07:32 PM
 #669

For amdgpu I use ohgodatool and rocm-SMI
http://nerdralph.blogspot.ca/2017/03/amdgpu-pro-1660-on-ubuntu-kernel-4105.html

When you're ready to come back to the deep end of the pool we'll still be here. :-)


on the rx470 in question I ended up hard-coding mem/core clocks & voltages in the BIOS so no tool is needed to change speed, happy with it on Win10 LTSB for now but may try linux again, thanks Smiley

question: on your blog, in the amdgpu screenshot it shows that 2100 MHz is the max, why is overclocking limited in linux? (or do you just not care becuause 2100MHz happens to be perfect for eth using your tightened timings?)

If you look back in this thread you'll see where I posted about hitting a limit of 2100 with this particular card.  Lots of Linux users have cards (usually Rx 480) running up to 2200 or even higher.
kilo17 (OP)
Legendary
*
Offline Offline

Activity: 980
Merit: 1001

aka "whocares"


View Profile
April 28, 2017, 09:40:02 AM
 #670

Sorry I have been away for a while.  Will read through the poss and see if I need to update the OP.

Bitcoin Will Only Succeed If The Community That Supports It Gets Support - Support Home Miners & Mining
OhGodAGirl
Full Member
***
Offline Offline

Activity: 199
Merit: 108

Look, I'm really not that interesting. Promise.


View Profile WWW
April 28, 2017, 10:16:34 AM
 #671

For amdgpu I use ohgodatool and rocm-SMI
http://nerdralph.blogspot.ca/2017/03/amdgpu-pro-1660-on-ubuntu-kernel-4105.html

When you're ready to come back to the deep end of the pool we'll still be here. :-)


on the rx470 in question I ended up hard-coding mem/core clocks & voltages in the BIOS so no tool is needed to change speed, happy with it on Win10 LTSB for now but may try linux again, thanks Smiley

question: on your blog, in the amdgpu screenshot it shows that 2100 MHz is the max, why is overclocking limited in linux? (or do you just not care becuause 2100MHz happens to be perfect for eth using your tightened timings?)

If you look back in this thread you'll see where I posted about hitting a limit of 2100 with this particular card.  Lots of Linux users have cards (usually Rx 480) running up to 2200 or even higher.


I've been a little busy with work, but just wanted to say + thank you for setting up a CGI wrapper online, to make it accessible for all. This is amazing work. You're a real gift to the community, I hope you kjnow that, nerdralph!
doktor83
Hero Member
*****
Offline Offline

Activity: 2520
Merit: 626


View Profile WWW
May 03, 2017, 10:03:49 PM
 #672

Ok, do GPU Mem errors shown in HWinfo affect the number of found shares ?
I mean more errors mean less found solutions, or it has nothing to do with it?

SRBMiner-MULTI thread - HERE
http://www.srbminer.com
PontiacGTX
Member
**
Offline Offline

Activity: 71
Merit: 10


View Profile
May 06, 2017, 02:43:29 PM
 #673

what could be a stable latency for H5GC4H24AJR that cna OC fine at 1900MHz?

▒ NOW token ▒ by ChangeNOW ▒ Get the WIN! ▒
ChangeNOW - an instant Non-custodial Exchange Service  (( changenow.io ))
Whitepaper  ▓  Telegram  ▓  Twitter  ▓  Facebook  ▓  Medium  ▓  Reddit  ▓  Bounty Thread
lexele
Full Member
***
Offline Offline

Activity: 190
Merit: 100


View Profile
May 06, 2017, 05:07:09 PM
 #674

Ok, do GPU Mem errors shown in HWinfo affect the number of found shares ?
I mean more errors mean less found solutions, or it has nothing to do with it?

 In my experience,If error in the millions, yes. usally you can see share rate decline and eventuelly a crash a the end.
doktor83
Hero Member
*****
Offline Offline

Activity: 2520
Merit: 626


View Profile WWW
May 06, 2017, 05:12:22 PM
 #675

Ok, do GPU Mem errors shown in HWinfo affect the number of found shares ?
I mean more errors mean less found solutions, or it has nothing to do with it?

 In my experience,If error in the millions, yes. usally you can see share rate decline and eventuelly a crash a the end.


i have millions of errors, and very rarely rejected share. So it looks like its not really a big problem Smiley

SRBMiner-MULTI thread - HERE
http://www.srbminer.com
mankeen
Sr. Member
****
Offline Offline

Activity: 644
Merit: 292


View Profile
May 06, 2017, 06:34:32 PM
 #676

What strap has given you the best hash rate with the rx480?
nerdralph
Sr. Member
****
Offline Offline

Activity: 588
Merit: 251


View Profile
May 08, 2017, 01:24:47 PM
 #677

What strap has given you the best hash rate with the rx480?

It depends.  The answers are in this thread if you take the time to read it.
nerdralph
Sr. Member
****
Offline Offline

Activity: 588
Merit: 251


View Profile
May 08, 2017, 01:27:11 PM
 #678

I've been a little busy with work, but just wanted to say + thank you for setting up a CGI wrapper online, to make it accessible for all. This is amazing work. You're a real gift to the community, I hope you kjnow that, nerdralph!

Well, depending on who you ask, I'm also a big asshole. :-)
bughatti
Full Member
***
Offline Offline

Activity: 154
Merit: 100


View Profile
May 10, 2017, 06:55:07 PM
 #679

So I know I have a ways to go, but I spent last night going through this thread and a few articles on other pages off of this thread.  I know this thread was created to help people understand so that is what I am trying to do, i am not hear to ask others to create my values.  Few questions to start off. So understanding some of the posts it seems tightening the timings is more memory specific than card specific.  So I can create a value from a rx 480 sapphire with hynix and it could work on a gigabyte rx480 with hynix?  When looking at timings for 4gb vs 8gb, changes can be made in different places.  For example on my xfx rx 480 8gb with samsung I see only samsung timings but in my sapphire rx480 nitro+ 4gb, I see places I can change for either samsung, or elpida/hynix.  I am using srbpolaris / polaris and even with my 570 I see changes on 4gb cards being made at the 2: section for elpida/hynix vs 1: for samsung.  I know that is just strap copying, but gotta start at basics.  In srbpolaris (I know most of you guru's use linux, and it is not my forte, but will work on learning) I am able to analyze a strap and I am seeing alot of sections I can change.  After reading some more through the thread, I see that it is not always a specific set of values for every card you can change, seems I need to learn the cheat sheet for what each value does.  It seems FAW is one that can always be set to 0 but I dont see 32AW (at leasnt not in srbpolaris). TRP can be changed it seems pretty consistently throughout all cards?  Also I am unsure how to decode the MISC.  I could be using the wrong tool, and if so I just need to spend more time on what else I can use.  Any guidance you all can provide would be really helpful and appreciated. 

Sorry if my thoughts seem random, I was trying to remember all I had thought about last night while reading!
deadsix
Hero Member
*****
Offline Offline

Activity: 751
Merit: 517


Fail to plan, and you plan to fail.


View Profile
May 10, 2017, 08:29:48 PM
 #680

So I know I have a ways to go, but I spent last night going through this thread and a few articles on other pages off of this thread.  I know this thread was created to help people understand so that is what I am trying to do, i am not hear to ask others to create my values.  Few questions to start off. So understanding some of the posts it seems tightening the timings is more memory specific than card specific.  So I can create a value from a rx 480 sapphire with hynix and it could work on a gigabyte rx480 with hynix? 

Yes, for instance a 4GB memory strap for hynix will work across brands like Sapphire, HIS, MSI as long as the card has the same 4GB hynix chips for memory.

Quote
When looking at timings for 4gb vs 8gb, changes can be made in different places.  For example on my xfx rx 480 8gb with samsung I see only samsung timings but in my sapphire rx480 nitro+ 4gb, I see places I can change for either samsung, or elpida/hynix. 

Most 570 cards come with multiple timing sets even though they have only one kind of memory. In such cases, only mod the timing set that is relevent to the kind of momory the card is using (identifying what the memory is by looking at the stock strap comes with xexperience)

Quote
I am using srbpolaris / polaris and even with my 570 I see changes on 4gb cards being made at the 2: section for elpida/hynix vs 1: for samsung.  I know that is just strap copying, but gotta start at basics.  In srbpolaris (I know most of you guru's use linux, and it is not my forte, but will work on learning) I am able to analyze a strap and I am seeing alot of sections I can change.  After reading some more through the thread, I see that it is not always a specific set of values for every card you can change, seems I need to learn the cheat sheet for what each value does.  It seems FAW is one that can always be set to 0 but I dont see 32AW (at leasnt not in srbpolaris). TRP can be changed it seems pretty consistently throughout all cards?  Also I am unsure how to decode the MISC.  I could be using the wrong tool, and if so I just need to spend more time on what else I can use.  Any guidance you all can provide would be really helpful and appreciated. 

Use nerdraplhs tool linked earlier in this thread, he posted it as a simple weblink that you can easily access.

Ethereum/Zcash/Monero Mining Bangalore https://bitcointalk.org/index.php?topic=1703592
Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 [34] 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 »
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!