Hat das jemals funktioniert? Beim A8N32 mit Toledo meines Wissens nach nicht....
[Sammelthread] Sockel 939 / 940 / 754 - Thread
- Ersteller The Sandman
- Erstellt am
Habe ich selbst noch nie probiert, weil ich normalerweise wegen der komischen Teiler immer den HT 1:1 zum Ram fahre. Sonst wird man wahnsinnig 
Mit E4 San Diego und E6 Venice sollte es aber definitiv gehen.
www.anandtech.com
Mit E4 San Diego und E6 Venice sollte es aber definitiv gehen.Athlon 64 Revision E: Unofficial DDR500 Support
Sparksnl
Enthusiast
I tested some more and I now noticed negative scaling with pi32m and adding more voltage with at least one 512MB dimm.
Sparksnl
Enthusiast
In between result. 3.41v
Let’s hope it is enough for tommorrow.
Let’s hope it is enough for tommorrow.
digitalbath
Mr. Oldschool-Bios
A good result in my opinion. I need to test my 512MB AT6 dimms with my DFI NF4 board. I see you raised Trfc and the Vcore voltage. I guess it was necessary to get that result, right?
Did you try to raise Trc? something to 9 or 11?
Did you try to raise Trc? something to 9 or 11?
Sparksnl
Enthusiast
Voltcore was just to be sure (this is not my best 4000+ BY batch). I tried trc as well but it was not the most important factor so far.
I still need to do much testing. But this is already much better.
I used ERam and prio very high as well.
I needed to at least do this because tomorrow I will run my fx-57 on cold.
Let’s see what pi 32m can do then.
I still need to do much testing. But this is already much better.
I used ERam and prio very high as well.
I needed to at least do this because tomorrow I will run my fx-57 on cold.
Let’s see what pi 32m can do then.
trc 7 is quite tight, while TRFC 15 looks a bit loose, at least from memory. Have you tried either 7/12 or something like 9/13?
Also the async latency and read preamble had a huge influence on stability in my tests. I’m not sure how I set these though. You might be able to find good settings in old threads on Overclock.net forums or XS. Drive strength I remember setting to either 6 or 8 for Winbond and 5 or 7 for TCCD.
EDIT:
Take this with a grain of salt. Been ages i tweaked on DFI NF4... It's about time to start again
Also the async latency and read preamble had a huge influence on stability in my tests. I’m not sure how I set these though. You might be able to find good settings in old threads on Overclock.net forums or XS. Drive strength I remember setting to either 6 or 8 for Winbond and 5 or 7 for TCCD.
EDIT:
Take this with a grain of salt. Been ages i tweaked on DFI NF4... It's about time to start again
Zuletzt bearbeitet:
Sparksnl
Enthusiast
In my own testing Drive strength with 7 was best for 256’s. Not sure if that is best for 512’s as well.
I did try 9/13. That did nothing for stability.
7/12 would seem unlikely to be working great because it needs 7-9/14 or 15.
I need trfc to be at least 14 to make it work. Well at least for the last 2 MHz or so.
I can do more testing with read preamble.
I did try 9/13. That did nothing for stability.
7/12 would seem unlikely to be working great because it needs 7-9/14 or 15.
I need trfc to be at least 14 to make it work. Well at least for the last 2 MHz or so.
I can do more testing with read preamble.
MusicIsMyLife
Urgestein
Ich habe das genau 1x auf einem DFI-Board verwendet: Dem RDX 200.
Also ja, es hat technisch gesehen funktioniert. Jedoch für Low-Latency-RAM sinnlos, da man mit 1:1 nie in irgendwelche Limite gelaufen ist (weder CPU noch Board).
Sparksnl
Enthusiast
New personal FX-57 CPU frequency record (4072MHz):
valid.x86.fr
About PI 32M. Sadly my memory needed to be stable 1 MHz more to be able to run faster (18.5x CPU divider). So this is it for now. More binning to do (3868MHz):
PI 1M for now.
And I need to optimize OS and my PI running. But at least I now know what it can do.
PI32M might do 3.9GHz as well. And I expect 3910-3920MHz to be stable.
AMD Athlon 64 FX-57 @ 4072.09 MHz - CPU-Z VALIDATOR
[ke7i1x] Validated Dump by Sparks.nl (2024-11-17 16:20:38) - MB: DFI LP UT NF4 Expert - RAM: 600 MB
valid.x86.fr
About PI 32M. Sadly my memory needed to be stable 1 MHz more to be able to run faster (18.5x CPU divider). So this is it for now. More binning to do (3868MHz):
PI 1M for now.
And I need to optimize OS and my PI running. But at least I now know what it can do.
PI32M might do 3.9GHz as well. And I expect 3910-3920MHz to be stable.
Zuletzt bearbeitet:
Richtig starkes Ergebnis, vorallem 32M stabil. Sieht man sehr selten, dass man 20 Minuten den Benchmark laufen lässt. Da muss man immer wachsam sein, wenn LN2 nachgekippt werden muss.
Im HWBot discord gibts mittlerweile eine Kategorie mit diversen Images. Das von dir genannte sollte auch dabei sein.
Tapakah XP und Barbonet XP sind jedenfalls dabei…
Zuletzt bearbeitet:
Ähnliche Themen
