Jump to content
HWBOT Community Forums
Noxinite

AM3 RAM OC?

Recommended Posts

Just tested another 955 CPU and imc better, on water CPU... mem 940 6-6-5-16-11 , 1.77v. The mem seem to scale dram volts a little more, compared to my other chip, on water the core and uncore not as strong tho :(

Edited by Bullant
  • Like 1

Share this post


Link to post
Share on other sites
On 9/18/2018 at 11:47 PM, Bullant said:

Hey my board doing that fail boot thing too. So fun to pass pi then fail to boot when you restart and need to clear and reapply :( Saber tooth 

It's a complete POS board, is terrible for maximising CPU's on cold, for some reason it limited me be nearly 800MHz max clock (FX-8350) on both boards I tested compared to a CHV.

  • Like 1

Share this post


Link to post
Share on other sites

Did a little eff testing today on the 955 amd, is ran on single stage -30 and memory are 1.83v set bios....will run on ln2 next time refill, should see good boost when push the uncore more

 

1.png

  • Like 2

Share this post


Link to post
Share on other sites

Was testing PSC other night on 1090t a fail run on loop 5.... the couple runs I tried.The cpu was -50 and memory -30, memory will tighten up more but seems imc related issue on the freq side. Given little more time should improve, will try more cpus and boards as I find them within my budget  ahahah

 

1z6tu87.png

Edited by Bullant

Share this post


Link to post
Share on other sites
On 9/27/2018 at 12:03 PM, Bullant said:

Did a little eff testing today on the 955 amd, is ran on single stage -30 and memory are 1.83v set bios....will run on ln2 next time refill, should see good boost when push the uncore more

 

1.png

I'm curious to hear why your tRAS and tRC are so loose - is that what it took for your memory to pass 32M @ 940, or is it just faster? I was running 18-11 recently, but that was around 880 MHz.

Share this post


Link to post
Share on other sites

Bit of both, that imc not the greatest and I dont see real big gain dropping lower then what I am, so I rather the extra MHz over dropping them little more 

 

Even when I can drop down to the "11" it's worse for me, or not worth it,  I try to use all that I see as a guide only and run what is faster for me. Was like haswell how some others see gains in couple of tighter subs , I didn't. I found the same with PSC on sandy and ivy, below tras 21, like 18 was bad for me 

Ran this one much the same tho little more mem MHz .Its the only really hyper I've tested or used, so I can't compare mem 

1090t.png

  • Like 1
  • Thanks 1

Share this post


Link to post
Share on other sites

Nice job @Bullant, AM3 looks like a lot of fun! Long time not benching really want to benching something. And i have same experience, tras 21 is faster on sandy than 18 or 15, tested on samsung and psc.

  • Like 2

Share this post


Link to post
Share on other sites
2 hours ago, speed.fastest said:

Nice job @Bullant, AM3 looks like a lot of fun! Long time not benching really want to benching something. And i have same experience, tras 21 is faster on sandy than 18 or 15, tested on samsung and psc.

You should do it Speed, its slow but same for everyone lol Its also cheap is what I like ahah

 

 

  • Like 2
  • Thanks 1

Share this post


Link to post
Share on other sites

Nice Bull.

I never figured out why my score was so slow. When I rebenched for TC I tightened RAS and RP and knocked off 6-8s, but still not good. :/

  • Like 2

Share this post


Link to post
Share on other sites
20 minutes ago, Noxinite said:

Nice Bull.

I never figured out why my score was so slow. When I rebenched for TC I tightened RAS and RP and knocked off 6-8s, but still not good. :/

Board. Most people were benching on non-efficient boards, some even on an awful boards for Deneb, e.g. Sabertooth or M5A99X.

I was expecting some LN2 scores from you, though :P

  • Like 1

Share this post


Link to post
Share on other sites

Is it required to only bench AM3 on Giga boards then?

I tried some LN2 for TC, but only on 939 32M. It didn't go well, so I gave up. :(

  • Like 1

Share this post


Link to post
Share on other sites

No, probably that mean 790FX/890FX is needed for Deneb/Thuban. 990FX/990X for FX. Probably its true because 990FX not good for high HTT on older CPU.

15 hours ago, I.nfraR.ed said:

To be honest, looking at country cup scores, @Bullant is the only one with a nice efficiency in 32M.

Hopefully can comeback soon

Edited by speed.fastest
  • Like 1

Share this post


Link to post
Share on other sites
3 hours ago, Noxinite said:

Is it required to only bench AM3 on Giga boards then?

I tried some LN2 for TC, but only on 939 32M. It didn't go well, so I gave up. :(

Practically - yes. But not every board and not every revision. 890 ud5, ud7 are good, some of the 9xx boards too, but I think it greatly depends on revision and we know Gigabyte has ton of them... My 990FX-UD3 rev 3 is the worst board I've ever had. From Asus - Crosshair III. Not sure about other brands. Crosshair IV isn't bad, but still slower.

The board I am using is very tricky with awful vdroop and only good for single-threaded benches, not really suitable for multithread like Cinebench and wrpime.

  • Like 1
  • Confused 1

Share this post


Link to post
Share on other sites

Tested a 960T and 990Fx UD3 tonight, cpu SS only about -25 because no Paste lol....is why couldn't get any higher NB Freq. Hypers @1.85v...will run this again later on the 890FX UD7

 

This board is rev 1.1

1.png
 

1.png

Edited by Bullant
  • Like 1

Share this post


Link to post
Share on other sites
2 hours ago, Bullant said:

Tested a 960T and 990Fx UD3 tonight, cpu SS only about -25 because no Paste lol....is why couldn't get any higher NB Freq. Hypers @1.85v...will run this again later on the 890FX UD7

 

 

890 board is the ticket for Thuban and Deneb. ;) 

  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, Mr.Scott said:

890 board is the ticket for Thuban and Deneb. ;) 

Older revisions of 9xx giga boards are surprisingly efficient (non-uefi bios). Not sure if all of them, but at least some.

  • Like 2
  • Thanks 2

Share this post


Link to post
Share on other sites
7 hours ago, Bullant said:

Tested a 960T and 990Fx UD3 tonight, cpu SS only about -25 because no Paste lol....is why couldn't get any higher NB Freq. Hypers @1.85v...will run this again later on the 890FX UD7

This board is rev 1.1

I actually have that board, it was efficient for me too IIRC. But I managed to get plasticene in one of the ram slots, so now it doesn't clock ram very well.

  • Like 2

Share this post


Link to post
Share on other sites
8 hours ago, I.nfraR.ed said:

I guess it is a non-uefi board.

It does have uefi option in bios, I seen it...I've not touched it.This is little new to me, if it has that option in bios is it a uefi board? I'm at work now so can't look at it to post a pic, when home I'll  look again to be sure....unless you know what it is? 

6 hours ago, Mr.Scott said:

890 board is the ticket for Thuban and Deneb. ;) 

Yeah 890 seems good, thanks mate.....I'll run it up again on my 890 ud7 soon 

Edited by Bullant

Share this post


Link to post
Share on other sites

They have the old trusty Award bios interface, while newer revisions have the new flashy, slow and buggy graphical UI where you can also use mouse.

  • Like 3

Share this post


Link to post
Share on other sites

My 990Xa-ud5 outperform CrIV in all 2d benchmarks. The gap is not so big but ud5 shows that in every benchmarks where I saw.

I guess some memory settings can be the main reason. 

Edited by Alpi
  • Sad 1

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×