[Sammelthread] Sockel 939 / 940 / 754 - Thread

[McZonk]

War as damals eigentlich teildefekter Salvage oder Bedienung hoher Nachfrage?

 

[Raiden Zero]

War as damals eigentlich teildefekter Salvage oder Bedienung hoher Nachfrage?

Würde vermuten eher ersteres. Die Venice wurden ja auch in China produziert. Da gabs sicher insgesamt genügend Kapazitäten.

Wissen wird das aber nur AMD selbst.

 

[Sparksnl]

Nachdem ich im Marktplatz-Thread ja schon den Verdacht geäußert hatte, dass CAB1E Venice vermutlich eigentlich Opterons mit teil-deaktiviertem Cache sind:

Anhang anzeigen 881469

Von links nach rechts:
Opteron 280 - CAB1E Venice - NBBWE Venice

Hab jetzt gerade keinen geköpften 1MB Singlecore zur Hand, die sehen aber genauso wie der CAB1E aus.

Wobei das grundsätzlich ja klar war, da CAB1E immer 1MB Cache bedeutet.

Edit: Hab gerade mal geschaut. 3800+ gibts auch als CAB1E. Das könnte richtig interessant sein, wenn’s bis ca. 310BCLK bei passender Temp geht.

Und ich hab auch noch nen interessanten 3700+ im Zulauf. 0630CPMW

I often thought of it as a pissibility that my golden CAB1E is a half cache 1MB CPU. The one thing I do wonder. Venice shows E6, 1MB’s show E4.

Edit: After running FSB tests for weeks I will do some sub zero tests soon. Starting with my Opteron 154 and a 3500+ LBBWE which had an extremely high FSB on air and boots 3GHz watercooled.

 

[Raiden Zero]

The one thing I do wonder. Venice shows E6, 1MB’s show E4.

This I also don’t really understand. Afaik E6 are usually X2 with one core disabled so maybe AMD used E6 for partially disabled CPUs in general?
So basically 512KB X2 E6 used to be 1MB X2 and 512KB/1MB single core E6 either X2 1MB, 512KB or single core 1MB.

I’ll just delid one of my E6 Venice with BW real quick and have a look.

Edit:
So at least for BW E6 Venice my guess was incorrect. Looks just like a regular Venice E4.

 

[Sparksnl]

I just ran my Opteron 154 on cold. This was a cursed run. My board/cpu combi started slowing down my post when I did not put the machine (SS) on yet. I hoped that after a bios reset and turning my cooling on, it worked properly, but it did not. So booting in 5-10min was not workable. But: it did post and boot at -60C. Even when turning my SS off and running at +20C did not help.

 

[Raiden Zero]

Sounds strange. Did you test a different CPU/RAM? For me that sounds more like a storage/external device problem.

 

[Sparksnl]

I will test it later. I had another SS session planned right after this.
So far: 3500+ and dr expert board 2 run fine on the same bios version and memory.
Runs 3400-3550MHz so far. Needs more cold, -60C is not enough. So that makes a nice nr2 3500+…

 

[Raiden Zero]

3550 sounds good. Is this benchable or only for values?
BW I guess as it still runs at -60?!

 

[Sparksnl]

Benchable MHz range is 3380-3450MHz I think.
For example: WPrime 32M does 3421-3440MHz and PI 1M did 3421MHz.

And I have a 3.2GHz bootable cpu (cab2e) as well somewhere…

Edit: 3550MHz is validation only.
Edit 2: max FSB at -60C is 370MHz. So -80C can probably still run.

 

[Raiden Zero]

-80c might give you at least additional 100MHz I‘d say.

Tested mine a little further. 3330 I can run at 1.47V watercooled but after that it really needs a huge voltage bump. 32M looks like 3300 at 1.52v is max watercooled.

 

[Sparksnl]

Is that your 3500+?

I just tested board 1 with my Opteron 154 combi. Same hardware and OS: runs fine now… watercooled that is.
I might want to test that board 1 with a known good cpu. So I know for sure it is Ok for sub zero. I had some issues last time as well. This time I lost USB again subzero and it slowed down.
I will start with erasing/writing that bios again.

Oh and I noticed that my dr Expert shows negative temps in BIOS hardware monitoring. At least up to -40C for cpu.

Edit: I found the issue of my ruined run. Defective usb mouse. As soon as I plug it it, posting is slow.

 

[McZonk]

CAB1E never gets old (nor disappoints) 305 x 10 sind unter Luft mit 1,472 Volt (1,500 BIOS) auch primestable.

AMD Athlon 64 3200+ @ 3247.01 MHz - CPU-Z VALIDATOR

[k3zpf9] Validated Dump by McZonk (2023-04-29 14:49:44) - MB: DFI LP UT NF4 Expert - RAM: 2048 MB

valid.x86.fr

 

[Sparksnl]

They are in, no matched set indeed.

 

[Raiden Zero]

Still these should clock fairly well. My best overclockers even on S940 are CAB1E. BB can be on par but same as on 939 BW will suck.
Which is why there are no BW 254 or 256 for example.
On 15x cold bug shouldn’t matter as my B1 on 940 usually do 260+

 

[Sparksnl]

Yes, B1 seems best so far for air. BB is Ok for air as well.

I tested my Opteron 154 on cold yesterday. That is a B1 as well. It has a max FSB of 395MHz on air.
On cold (-60C, Single Stage) I have a maximum of 250MHz FSB depending of which multi I use. 5x and 12x give 250. 13x - 14x give 245MHz FSB. 14x245 needed 1.7v to boot, any lower voltage gave a CBB (Cold Bug Boot), no full boot.
I could post 14x200 @1.35v without any issue, so no Cold Bug for posting the pc.
So this one needs less cold. I would guess a range of -30C to -40C.

 

[Raiden Zero]

-60c might be a little to much cold as you already said. Mine does -44c max, but it’s still too much for low multiplier B1 CPUs.
———

Hab heute ein paar neue CPUs bekommen. Unter anderem ein 4200+ LCBQE 0723VPMW. Die CPU scheint leider beschädigt zu sein.
Das DFI bootet mit der CPU nur nach vormaligen CMOS Reset. Selbst optimized defaults gehen nicht, das Board bleibt dann mit F2 hängen. Komme zwar ins Windows nach cmos reset aber beim reboot dann wieder das gleiche. Ganz merkwürdig.
Auf den ASRock startet die CPU erst gar nicht.

Jemand ne Idee?

 

[Sparksnl]

-60c might be a little to much cold as you already said. Mine does -44c max, but it’s still too much for low multiplier B1 CPUs.
———

Hab heute ein paar neue CPUs bekommen. Unter anderem ein 4200+ LCBQE 0723VPMW. Die CPU scheint leider beschädigt zu sein.
Das DFI bootet mit der CPU nur nach vormaligen CMOS Reset. Selbst optimized defaults gehen nicht, das Board bleibt dann mit F2 hängen. Komme zwar ins Windows nach cmos reset aber beim reboot dann wieder das gleiche. Ganz merkwürdig.
Auf den ASRock startet die CPU erst gar nicht.

Jemand ne Idee?

Maybe check a different bios version?
Or try different manual settings with some still on auto like voltages or memory settings.
I have 3 4200+ LCBQE 0722XPMW. These are problem free so far. All 3 do a 3.1-3.2GHz boot.

Edit: I do know that after running a Clawhammer and posting an E3, E4 or E6 I have to reset my bios and make a new bios profile. It won’t work without doing this. So do not load a saved bios profile in that case.

 

[Raiden Zero]

Maybe check a different bios version?
Or try different manual settings with some still on auto like voltages or memory settings.
I have 3 4200+ LCBQE 0722XPMW. These are problem free so far. All 3 do a 3.1-3.2GHz boot.

Edit: I do know that after running a Clawhammer and posting an E3, E4 or E6 I have to reset my bios and make a new bios profile. It won’t work without doing this. So do not load a saved bios profile in that case.

BIOS I might try but I don’t see it changing much as the same CPU won’t even post on my asrock.

I have other LCBQE to check but looks like a faulty CPU so far.

Right now I’m back to testing my already binned CPUs on CFX3200 just to see if they might perform any better compared to the NF4 one.


Meine 4000+ hab ich durch, wird Zeit auszusortieren.
Ausgangspunkt war boot bei 1.55V + 1,4% BIOS (1.55Vc real) mit 250x12:

KAB2E 0549BPMW: no boot
LCBBE 0640BPMW: 3050MHz 1M
LCB9E 0644XPMW: 3120MHz 1M
KAB1E 0631GPBW: 3230MHz 1M
KAB1E 0631APMW: 3400MHz 1M



Auf dem Screen fehlen ein paar MHz, da entweder meine UTT oder der Mem-Controller bei höherem Takt schlapp macht. Mit 5/6 Multi laufen auch 3425MHz.

Bewegt sich beim Vergleich CFX3200 vs Ultra-D alles im Bereich +/- 10MHz, den 0631APMW hab ich geköpft. Da bin ich wirklich gespannt, ob die CPU mit Chiller die 3600MHz 1M knackt.
Lasse gerade 32M laufen, da die CPU vermutlich anstelle des 4800+ ins Retro-System kommt. Sieht aus, als wären 3200MHz bei 1.39V drin.

 

[käfdox]

Schöne CPU

 

[Sparksnl]

That is a nice cpu. I actually have a good 4000+ air cpu which is only slightly worse. It is a KAB1E 0631GPCW. Also 0631.
So far for GPCW not being good…

Edit: Best one (for me) remains my 3500+. 3.4GHz boot @1.44v.
That 3500+ must have been a mistake from AMD side 😉.
3629MHz running it outside in my garden with a 8C outside temp.

 

[Raiden Zero]

Now that I think of it: I’m pretty sure I got that one from @TAGG back then on eBay.

3.4g bootable at that voltage sure is impressive. Above 4000+ is capable of roughly the same (1.47V) but that CPU is delidded which allowed for up to 100MHz more.

 

[Sparksnl]

Both my 3500+ and my kab1e 4000+ are not delided. I might give it a go for that 4000+ since I have a golden 4000+ anyway. And actually,.. I have that 3.9 monser cpu, a 3.3 not delided and several 3.2’s. So in this case I can safely take a risk.

Did you run them direct die at max clocks?

 

[Raiden Zero]

Did you run them direct die at max clocks?

Yes. Doesn’t make much of a difference if you run it direct die or just replace the TIM and remount the HS properly.
In most cases it seems like bad HS mounting is the problem.

 

[Sparksnl]

@Raiden Zero Can you give some photo's and writing about deliding those cpu's. I notice it is too long ago for me.
I delided a CPU a few years ago but still it seems as it I do it the wrong way.

 

[Raiden Zero]

I’ll see what I can do tomorrow. Actually delidding those is fairly easy. I simply use a razor blade to cut in between the CPUs PCB and the HS. Just make sure not to cut any deeper than 3-4mm and try to stay close to the HS (and away from the PCB). At the corners of the HS you can go a little deeper as there is more space.

If needed you can use a thicker knife to gently lift the IHS off once you’ve cut all the silicone.

Did this to my CABYE 0528GPMW Opteron 152 today. Before by best for 1M under SS was 3450, now I’m at 3530-3550. I really need to test this one with ln2, so far I can go up to 335MHz which is the limit on my CFX. So there’s quite a lot of room left using x13 multi.

 

[bschicht86]

KAB1E 0631APMW: 3400MHz 1M

Das Teil habe ich auch, welches mit 1,3V und 3GHz bootet.

 

[stunned_guy]

4000+ 0631APMW war das letzte Stepping, welches weltrekordfähig war und eines der letzten E4 San Diegos.
Das war nochmal ein Hype nach den ganzen 3700+ 0547GPMW und 0614BPMW Highlights.

 

[Blackpitty]

ich hab den hier rumfliegen

 

[stunned_guy]

Auch nicht schlecht War kurz vor den 0631 APMW eine gute Serie gewesen. 0626FPMW. Allerdings nur bei den 4000+. Bei den 3700+ gab es nur Krücken.

 

[Blackpitty]

den gleichen wie meiner gibts bei Ebay grade für um die 20€(zu teuer)