Phenom 9500 w/ MSI K9A2 Platinum

[macci]

just to clarify couple things as it seems that someone had problems locating the point of the post:

1) with the latest BIOS the MSI board runs fine with 1066 memclk mode, even with 800SPD dimms

2) NB clocks are unlocked with the BE (I think this is the same with non BE too though) same goes to HT, same goes to Voltage too actually (!)

3) there is nothing "unreachable" about these clocks, its just an example config and I am not maxing out the CPU here - just showing a combination of things (1066, high NB clock, OVCKed CPU and then Mem performance)

4) retail 9600 BE, no cherry picking - also as mentioned it was an easy clock, not pushing it

there is absolutely nothing special about those screenshots, everything can be set directly from the BIOS too

I just set it up yesterday evening and did a quick run and posted the result

I am sure it can run faster (we have already seen way higher clocks with BE)

overall I think its good news for the guys with the MSI board

[Swanie]

Yes thanks Macci that is good news. Keep the news coming and also KTE- without you guys me for one would not have known how to unlock the potential and possible bright future of the Phenom Chip.

[KTE]

vaughanster: Stay out of the thread unless you have something sensible to add. You need to read my very first post before replying and oddly, you sound exactly like one of the extremists already banned for attacking macci in the 3DMark thread. And yes, grow up.

Sammi: Your input is greatly appreciated as it shows exactly the things we've been asking for an complaining about not having. Again, please ignore the trolls.

i got my 9500 to 2.5ghz on 1.3v. so im guessing if i up the voltage to 1.35v i could maybe get to 2.7ghz

Try it. That's all I can say. Try for 2.5GHz and then 2.6GHz stability. This is the best way to go abut it. 1.45V should get 2.6GHz stable IMO. The three Phenoms I've yet oc'd all can reach 2.6GHz stable with that much or less VCore. BUt make sure you stress test for at least 8 hours continuous (left overnight is best) at any MHz before trying to higher on stability.

overall I think its good news for the guys with the MSI board

Agreed.

Yes thanks Macci that is good news. Keep the news coming and also KTE- without you guys me for one would not have known how to unlock the potential and possible bright future of the Phenom Chip.

Well, I need that new BIOS ASAP!

Everyone can add to the info we know, and many have added, not just me or macci. All is appreciated. That's how humans learn by working together.

[macci]

KTE no i havent, and mems are running @ 500mhz :P 2.3v for those but they can run 520 with those timings

but remember i did have very low nb speed on that test, i can rerun with 2ghz nb speed

yup, and also the CPU clock speed has a clear impact on Everest memory benchmark (Latency test also).

[macci]

hey KTE, check your PM

[Brother Esau]

[QUOTE=KTE;2677941]vaughanster: Stay out of the thread unless you have something sensible to add. You need to read my very first post before replying and oddly, you sound exactly like one of the extremists already banned for attacking macci in the 3DMark thread. And yes, grow up.


You know I was thinking the exact same thing last night KTE

[aGeoM]

Hi all

...

overall I think its good news for the guys with the MSI board

I'm using the performance bios P0D, and I can'd setup NB>9 without messing all reading in windows HT<200 and so on, it's possible for you to share the E version?

I would like to test that bios, if it can set NB multi > 9 with 9500 CPU.

Thanks.

*I'm sorry for my English.

All the best.

[KeZzZu]

hey KTE, check your PM

Thanks macci for information, ordering my be soon as possible when i get some extra cash (my job just ended ) some finding's regarding temps&OC

my max ht were 251 with boxed and +22 c' ambient temp and opening case + window (-8c' air directly to pc) idle did go down by 16 c' and my max ht wall raised to 261

someone can test it too?

ps: macci you got beta bios, erm could be pleasure down here ?

[KeZzZu]

Oh, Forgot tell this: ordered 2x3870 zerotherm models... also wc setup on it's way... only pump has arrived on time... thought well other parts on some where around finland on their way... all parts should be here before next friday

[KTE]

hey KTE, check your PM

Thanks as always macci.

Hi all

I'm using the performance bios P0D, and I can'd setup NB>9 without messing all reading in windows HT<200 and so on, it's possible for you to share the E version?

Hi. That's always been the case with every BIOS with locked Phenoms AFAIK. I've not seen any multi higher than 9x on any board with Phenom yet (without lower clock readings and no real NB change, same as CPU multi higher than stock would show). It looks like it's just not possible.

I would like to test that bios, if it can set NB multi > 9 with 9500 CPU.

ps: macci you got beta bios, erm could be pleasure down here ?

Both of you will have PMs soon.

my max ht were 251 with boxed and +22 c' ambient temp and opening case + window (-8c' air directly to pc) idle did go down by 16 c' and my max ht wall raised to 261

So you had a temperature problem?
I've never had any difference with Phenom @ 0℃ ambient or 26℃ ambient.

Whats max HT ref. on your board? And what about the Phenom (low multi)?

[KeZzZu]

KTE: have been using boxed cooler... @ 2.4ghz and full load temps are crawling near +65 c' imagine what it is at 2.7ghz

tested ht wall @ 8x 9x 10x. 11x multi has wall on 251 too but also temps are really bad so cant test higher than 10x multi

[Brother Esau]

Just out of curiosity why would you use that cheap cheesy cooler? Certainly cant be helping matters much on what appears to be challenged processors already

[KeZzZu]

Just out of curiosity why would you use that cheap cheesy cooler? Certainly cant be helping matters much on what appears to be challenged processors already

Becose every time when i buy new cpu, i will rape it with boxed cooler... (last time i did it on celeron L 440 3.8ghz run with near 1.7vcore... still alive that cpu )
And also it really show how cpu is affected by temp's and how it reacts... not sure if this phenom has more headroom for OC maybe 50mhz or maybe 300 mhz ? dunno yet.. it will be tested before next weekend

[blargman]

sry to get u all salivating eh, 1.24 bios i had was for k9a NOT the k9a2 ugh, my german needs polishing I have no secret sources, i go to the same german forum as the rest of ya

[KTE]

Thanks for clarifying blargman.

[Daveburt714]

Thanks as always macci.

Hi. That's always been the case with every BIOS with locked Phenoms AFAIK. I've not seen any multi higher than 9x on any board with Phenom yet (without lower clock readings and no real NB change, same as CPU multi higher than stock would show). It looks like it's just not possible.



Both of you will have PMs soon.

What No PM Luvin for the BURT?

My machine (Std. 9600) does the same thing when I set the NB Multi Higher than 9x. It will display it in AOD at the higher Multi, but it locks my HT Ref clock around 197 and won't allow me to adjust it... So, theres one confirmation anyway...

[KTE]

That behavior won't change Dave. Its happens on every Phenom (locked) we've seen yet.

However, which BIOS version are you using? Tried 1.13, P0C, P0D, 1.21, 1.22 etc yet?

[Brother Esau]

If you don't trust that for whatever reason, then wait till I or someone else here tests one.

Have you applied the bug patch? Your memory perf and latency is quite bad for 890 and those latencies.
But that maybe because you have a low NB clock.

Also, everyone should check this Phenom in-detail review out! I think this is the first one I really enjoyed for what they tried, dealt with, methods, investigated and wrote in regards.

1st Jan: http://www.lostcircuits.com/cpu/amd_phenom

Contents covered:
-V8, Penryn, Conroe, Pentium, 4x4, X2 (65nm/90nm) all compared.
-Int, FP, gaming benchmarks
-Effect of the AOD "magic button" across benches
-Effect of 1066 RAM over 800
-Power consumption in all states and vitaly when overclocking
-Power efficiency @ workloads
-Virtualization and TLB errata and patch (explanation)
-Effects of memory bandwidth
-Effects of NB clocks

Some things are a little inaccurate in there (one comment IIRC- will talk about it later) but the rest is v.good IMO. Pay special attention to how its shown that oc a lower VID/Amp/Volt binned retail CPU usually gets much lower CPU power consumption than one retailed at that Speed/Volts. This is the same with Pentium, Conroe, A64, K10 and Penryn across the board. Since MFG stability requires higher amps per core and thus higher TDPs/power than customer oc methods. This is why we with 9500/9600 were seeing much lower power consumption figures than to-be retail 9700 and 9900.

What a very thorough evaluation and enjoyable write up.....Props to the writer for sure

[justapost]

[b]Also, everyone should check this Phenom in-detail review out!

Interesting read, thanks for sharing.
NB-multi>9 work on the M3A but the performance drops. Have you tried it with FSB: DRAM 1:2?
See here.

[Daveburt714]

I've tried 1.0 and 1.13 (current)... If you have anything you feel is decent above 1.13 I'd be intersested.

My biggest problem seems to be the HT Ref. clock, it just doesn't seem to want to go much over 220 with any kind of stability at all. If I lower all multis to 5, I can get it up around 235, but even with everything set that low I still have stability problems.. Here's a screeny of the best stable I've gotten so far.



I know the V's are very low, but I've tried cranking them pretty high and if anything that seems to add to instability rather than settle things down...

Not trying to be a pest Bud, and I appreciate all you folks on this thread are doing new Phenom owners...

[KTE]

I agree there Campbell.

Oh and apparently, the "troll" earlier was Soldnermofo guy. Same IP.

NB-multi>9 work on the M3A but the performance drops.
See here.

Are you sure its working? That much has worked for everyone AFAIK, but can you show us an EVEREST/AOD reading alongside CPUZ?
Can you validate those CPUZ clocks?

I'll show you a way to test if your NB is really changing.
Throw the HT to the same multi as NB. If NB really is changing, the HT will be changed to the same speed. If not, the HT will not change above 9x which is because 9x is the real operating NB multi.

Try it. If its working, then you'll be able to validate with CPUZ and all software will be in sync (you can see macci's ss as a good example how). Also performance should'nt drop with higher NB but increase.

Have you tried it with FSB: DRAM 1:2?

Yep, I only ever ran DRAM 1:2. It doesn't change from stock 9x (upwards) and you can test EVEREST bandwidth to show you if it has changed. Higher NB definitely improves mem bandwidth/latency and L3 cache performance (see above macci's ss again). I've tried real (9x) 2160MHz and (8x) 1920MHz on NB at 2640MHz CPU to see the effects before.

[KTE]

Daveburt714: Have you tried putting CPU VID to 1.25 and then increasing CPU volts?
Your volts are way too low for 2.5GHz stability. You don't have CnQ enabled, do you? (should be disbaled if oc)

[Brother Esau]

KTE are these first generation boards sufficient enough for quad core overclocking? In other words what kind of board taxing do you have to put on it to overclock quad cores?

[Daveburt714]

Daveburt714: Have you tried putting CPU VID to 1.25 and then increasing CPU volts?
Your volts are way too low for 2.5GHz stability. You don't have CnQ enabled, do you? (should be disbaled if oc)

Yeah KTE, CnQ is disabled as well as Spread Spectrum... Those both caused OC problems on K8's so I assumed they would on K10 as well.

My board always boots with CPU Vid @ 1.2... I guess I could try and play with it in p-states to try and force it to 1.25.

The funny thing is, I actually played F.E.A.R. for 2 hours last night with those very settings, mainly to test stability, and it never even stuttered!!

Maybe I just have a wierd chip...

[justapost]

I'll show you a way to test if your NB is really changing.
Throw the HT to the same multi as NB. If NB really is changing, the HT will be changed to the same speed. If not, the HT will not change above 9x which is because 9x is the real operating NB multi.

Try it. If its working, then you'll be able to validate with CPUZ and all software will be in sync (you can see macci's ss as a good example how). Also performance should'nt drop with higher NB but increase.

Thx, only checked CPU-Z, so i did not realize the ref HT drop.
There is a msr register 0xc0010071. Bit 63:59 are the MaxNbFid. In my case 5 means 1800MHz max.