Phenom 9500 w/ MSI K9A2 Platinum

[KTE]

The +3GHz clocks are easier if the multi is higher and the board/CPU allows it. If my CPU allowed 11.5 multi it would get 266x11.5 and 267x11.5 max CPU-Z because it's the HT that messes everything up, not core speed.

I've worked around it and here are my findings:
I found out the issue is NOT the NB or CPU speed BUT the HT ref. speed.
The maximum for everyone I've seen is around 250-268 HT ref. Mine is 267MHz HT. ref.

Here I threw ALL speed including CPU/NB/HT to VERY low and only left HT ref. speed high. HT in total ran at 1x.

The HT ref. I tried with no volts added and max volts added. In both cases it goes to a maximum of 267MHz and then freezes the system. Here was 261HT, 265HT, 266HT and then 268HT up from 266HT when it froze.

http://img218.imageshack.us/img218/3...01sdsdssy6.png
http://img218.imageshack.us/img218/3930/asasaot0.png
http://img233.imageshack.us/img233/3180/dsc01908fs7.jpg
http://img233.imageshack.us/img233/3232/dsc01904ce8.jpg

I can boot 255HT ref. now stock volts, all cores.

All this looks like a CPU limit of HT ref. If anyone gets a chip that does 270MHz HT ref. all cores, they should be able to get +3GHz easy. We need to find out how to unlock higher multipliers...

267x13.5 looks especially good.

So at 2,2 GHZ you have 100VA power consumption in idle and 180VA under full load. With 2,6 GHz you have 120VA in idle and 200VA under full load.

AC. DC which is actual power consumption is lower, multiply by 0.83.

2.2GHz stock all volts, low HT gives ~177VAC max.
2.64GHz stock all volts, low HT gives ~200W VAC max.
2.7GHz max all volts, high HT, high NB gives ~330WVAC max.

You only had to change the Nb voltage +0,1V to archive those high NB and HT frequencies so i assume a similar power consumption.

NB needs those volts to achieve high speeds, yup.

This system runs more power efficient (GHz vise) if overclocked to 2,6GHz.

I think around 2.60GHz is most efficient.

As my 9500 still did not arrive and the M3A seems to be delayed till 12/12 may i ask for CnQ enabled (or lowest possible multis set) idle power consumtion at 200 vs 240 MHz HT?

What HT and NB total speed (not the reference clocks)? That'll make a major difference. Let me know that I'll try it and let you know.

HD 2600 XT's power consumption here in idle ~30W.

I've done my own measurements. Mine idles around ~25W and load is around ~50W max using ATi Tool.

So using an HD 2400 XT or pro instead should substract ~20W from the power usage resulting in 80W consumption for the whole system w/o CnQ.

I'll measure power draw at stock with all stock volts again and let you know what the DC W value is.

[informal]

KTE,seems you're the first one to find a limiting factor on your MSI board.Good going This is probably the case for all 790FX based boards(can we say Phenom has an issue with high HTT clocks and 790 chipsets and we need a high multi-aka BlackEdition- CPUs?).You did a great job for such a small amount of time .

OBR hinted while ago that he could reach much higher on HTT clock using a regular AM2 board(instead of 790 based one)

[KTE]

KTE,seems you're the first one to find a limiting factor on your MSI board.Good going This is probably the case for all 790FX based boards(can we say Phenom has an issue with high HTT clocks and 790 chipsets and we need a high multi-aka BlackEdition- CPUs?).You did a great job for such a small amount of time

This looks like a CPU limit rather than MB limit. The base clock of HT that has to multiplied has to remain low. You can see this and this exact reason why videos and higher clocked versions can only run faster due to higher CPU multiplers, but actually lower HT ref. than mine. HT total speed is not the limit, have done 2.4GHz on this but it's the HT ref. speed on mine and every other system I've seen. Limited on CPU at 250-268MHz.

OBR hinted while ago that he could reach much higher on HTT clock using a regular AM2 board(instead of 790 based one)

Are you sure he said this?
Maybe total HT speed can be higher on one of those boards but HT ref. I doubt very much is possible higher than these boards, quite honestly. These 790FX are the best boards for AM2+ as they support dozens of options and technologies no where near and better than older AM2 boards. Those boards can't even change FID/DID/VID on CPU and NB. These should be clocking best in all aspects by far.

Has someone reached HT ref. >265-268MHz with a 9500 on AM2 that you know of?

Hell yeah we need unlocked multi's! I think I've found a way for something here but not sure. Will try it out later...

[KTE]

the same as the multi-threaded bench.

you dont need to post screens...just the number is good.

Multi-thread: http://img218.imageshack.us/img218/9...nsinglewv0.png
Single-thread: http://img218.imageshack.us/img218/9...single2hn4.png

[Lightman]

Lightman,did you try using the EasyTune 5?SInce it seems that nguoisaigon managed 3Ghz with Phenom using this utility on your board.

This util doesn't like XP x64....

Besides AOD is fine for me, but my CPU is weakest from all I've seen so far .
Well I will hold on it till B3 probably .

[informal]

Are you sure he said this?
Maybe total HT speed can be higher on one of those boards but HT ref. I doubt very much is possible higher than these boards, quite honestly. These 790FX are the best boards for AM2+ as they support dozens of options and technologies no where near and better than older AM2 boards. Those boards can't even change FID/DID/VID on CPU and NB. These should be clocking best in all aspects by far.

Yep,i'm sure.He didn't post anything else since then,i hope he managed to OC it higher on regular AM2.
Here it is,i've found it :
http://www.xtremesystems.org/forums/...&postcount=290

"OBR:i am testing phenom on old nForce AM2 mobo and results are many better then on this Gigabyte! I am sure this Giga mobo has serious error in PCB design ..."

Has someone reached HT ref. >265-268MHz with a 9500 on AM2 that you know of?

Nope,no one that i am aware of.

Hell yeah we need unlocked multi's! I think I've found a way for something here but not sure. Will try it out later...

Yep,BlackBox will be a definite hit if it carries the same price and has at least a potential of 3Ghz on air(provided we avoid the HTT issue with unlocked multies,which i am sure you confirmed with your tests and finding out that HTT clock wall on regular Phenoms)

EDIT: One interesting link for you KTE:
http://www.msi-forum.de/board.php?boardid=52

Keep an eye on this forum .They are usually the first to get new/BETA bios versions for MSI boards(on the net!).The topic at the top of the forum is about your board .Check out all the pages(it's in German unfortunately ). 4 files attached,total.
Another thread(named "K9A2 Platinum - HT and PCI-E speeds in bios???":
http://forum.msi.com.tw/index.php?to...2045#msg842045
Another (maybe)useful post by an admin there:
http://www.msi-forum.de/thread.php?p...764#post296764

Quote:
Insert the function of HT3 Insert the function of HT3
Include the module of "AMD OverClock". Include module of the "AMD Over Clock."
Update the Promise-Rom, "PDC42819 BIOS v2.8.1030.0017." Update the Promise-Rom, "PDC42819 BIOS v2.8.1030.0017." <=[WTF they included HT3.0 only at this point?? ]

edit2: :
Found some new stuff for you:
link

Quote:
Original von Xervek Original by Xervek
einschlie&#223;lich der Performance-Optionen? Including performance options?

Nein, diese nat&#252;rlich nicht...die sind dem Performance-BIOS vorbehalten... No, of course not ... those are the Performance-BIOS reserved ... Freude

Weil wir gerade bei Performance-BIOS sind: Because we are at Performance-BIOS are:

<= Here he attached a so called performance BIOS,version A7376AMS.P0D.zip ( 536,43 KB)

Which BIOS version are you you running now?
Hope this helps at least a little bit.
Cheers

PS All translation done by google translate bot

[KTE]

EDIT: One interesting link for you KTE:
http://www.msi-forum.de/board.php?boardid=52

Keep an eye on this forum .They are usually the first to get new/BETA bios versions for MSI boards(on the net!).The topic at the top of the forum is about your board .Check out all the pages(it's in German unfortunately ). 4 files attached,total.
Another thread(named "K9A2 Platinum - HT and PCI-E speeds in bios???":
http://forum.msi.com.tw/index.php?to...2045#msg842045
Another (maybe)useful post by an admin there:
http://www.msi-forum.de/thread.php?p...764#post296764

edit2: :
Found some new stuff for you:
link

<= Here he attached a so called performance BIOS,version A7376AMS.P0D.zip ( 536,43 KB)

Which BIOS version are you you running now?
Hope this helps at least a little bit.
Cheers

PS All translation done by google translate bot

Thank you mate.

I've tested BIOS A7376AMS.100, A7376AMS.112 BETA, A7376AMS.113 BETA and A7376AMS.P0C. Running 113 right now.

Got dual channel and unganged mode working. Quick results.
http://img231.imageshack.us/img231/169/evrry3.png
http://img231.imageshack.us/img231/6112/evl1nj9.png
http://img231.imageshack.us/img231/6537/evc1vc5.png

I think my CPU has issues after 2.5GHz. From 1.8GHz to 2.55GHz it scales well with frequency but after that I hardly get performance increase. That's why I never concluded anything and didn't make a table comparison, neither with X2. It's too early to do that. That's with the last BIOS anyway, not tried the new ones.

Trying P0D in a little while.

[KTE]

As my 9500 still did not arrive and the M3A seems to be delayed till 12/12 may i ask for CnQ enabled (or lowest possible multis set) idle power consumtion at 200 vs 240 MHz HT?

CnQ enabled is like so: http://i222.photobucket.com/albums/d...4/CnQ-1100.png

That gave idle power of 96-98VAC.

Full load stock all. I measured again and it reached a maximum of 177VAC in P95.
Maximum of 168VAC in any application, which was 3DMark06 during Deep Freeze.

[nguoisaigon]

have you tried the A7376AMS.110 yet KTE? It has just been publish by MSI today. I think it will be better than .113

[KTE]

have you tried the A7376AMS.110 yet KTE? It has just been publish by MSI today. I think it will be better than .113

I'll try it today thanks.

BTW I forgot to mention to you: you should be able to get higher CPU-Z validation speeds.

Close all background processes. Have nothing open and open OverDrive 2.10
Then open CPU-Z

Go up 2MHz at a time and change HT reference clock by 2MHz each time - not higher. Do its slowly.

You should be able to get near 265-270MHz HT which at 11.5 multi is 3047-3105MHz.

Since the first BIOS, I haven't tried max on mine again but 2.92GHz was doable as is 266HT. But until it can bench 3GHz, I don't see the point in trying

[mongoled]

Great work KTE. Thanks for your efforts in bringing this information to us. Its good to see some people are making efforts to bring the platform with all its current attributes to XS so that a more informed decision can be made on the state of the Phenom platform.

Although not as great as we would have wished, I dont see the proposed doom of many nayslayers regarding this platform. People here have just been too spoilt in recent times, LOL.

I expect Phenom to improve alot, so that in six months time we will have a much clearer picture of the type of performance this CPU can bring. Its pretty obvious that the platform is not 'ready' for us enthusiasts, but this will be fixed in due course.

Its quite funny how some peeps have forgotten of many of the woes experienced in the past by other newly released technologies, as has been said by many, we have been spoilt.



Oh bandwidth has been exceeded on the pics as the rabid masses seek new informations

[KTE]

LOL! I just noticed you guys took up a whole 28GB bw in 3 days.

I'll rehost as soon as I can later. My apologies in advance.

At least pages load up quicker now... I think saaya did this on purpose=> j/k

If you really want to see the image before I rehost, then you can take the link and rehost it yourself through imageshack (etc) and it'll show up.

Thanks mongoled, appreciated. I agree on your analysis, I think it needs time. True we'd all love everything to work perfect right out of the box. We can't change whats happened and I've never been much of a dweller on the past. We know it was late and so on, we know it doesn't match Penryn and clock anywhere near as high on low multi, but my focus is on "how good is it? how bad is it? how efficient is it and its underlying technology as well as its potential? whats its idle/load temp/power consumption? wheres teh bottleneck? max stable on different speeds? what effect does different speeds have on perf/power/temps? is there a worthwhile product here in itself and in the platform? what perf can users expect day to day? what issues and capabilities are there with the CPU and platform? is it any better than an AM2 MB?" and so on.

I could've showed you 2-3 benches at 2.86GHz and ended with a to give you all false hopes of high MHz/perf possible easy, but that isn't reality. Fact is what is max stable on air and its perf is how you'll judge the CPU/Platform. So far I can't try that with this many issues, but 2.7GHz is certainly stable through OD.

[justapost]

The maximum for everyone I've seen is around 250-268 HT ref. Mine is 267MHz HT. ref.

Have you tried to find the max ref. HT the board can handle?

I can boot 255HT ref. now stock volts, all cores.

AC. DC which is actual power consumption is lower, multiply by 0.83.

Sorry non native english speaker here, that's what i had in mind.
I use W for DC and VA for AC.

2.2GHz stock all volts, low HT gives ~177VAC max.
2.64GHz stock all volts, low HT gives ~200W VAC max.
2.7GHz max all volts, high HT, high NB gives ~330WVAC max.
NB needs those volts to achieve high speeds, yup.
I think around 2.60GHz is most efficient.
What HT and NB total speed (not the reference clocks)? That'll make a major difference. Let me know that I'll try it and let you know.

VAC at 2,64 with high NT/HT would be cool.
Too bad you don't have a q6600 for comparison.

I've done my own measurements. Mine idles around ~25W and load is around ~50W max using ATi Tool.
I'll measure power draw at stock with all stock volts again and let you know what the DC W value is.

Great, how do you measure the card?

CnQ enabled is like so: http://i222.photobucket.com/albums/d...4/CnQ-1100.png
That gave idle power of 96-98VAC.

Full load stock all. I measured again and it reached a maximum of 177VAC in P95.
Maximum of 168VAC in any application, which was 3DMark06 during Deep Freeze.

Looking at your screenshot it seems CnQ does not reduce the HT multiplier.
You said HT speed has an big impact on consumption, so with a lower HT multiplier idle power should drop a little.
So i'm still confident a 9500 system will require less power than an q6600 in idle once CnQ (2.0 as in thw's review) works flawless.
Thank you for all the effort.

achim~

PS. 9500 finally on the road, should be here in a few hours. I'm really curious how far i will get in the M2A-VM, max. ref HT i could reach with an X2 3800+ EE was 273MHz so i'm confident to reach the cpu limit.

[KTE]

I flashed P0D BIOS (the supposed performance gaining BIOS), absolute nightmare! It just screws up stability and overclock potential although performance is a little better with it. Then I flashed an older one 110 and it's better but worse than 113 BETA. Cannot use OD and get decent MHz anymore, everything just freezes randomly.

Have you tried to find the max ref. HT the board can handle?

I can't because the CPU can't go further than 267MHz. And even that, not with the BIOS I'm using now, this one doesn't boot past 242MHz and OD just freezes system upon starting even at stock everything.

I'm trying 5000+ BE soon and so will try max HT ref.

VAC at 2,64 with high NT/HT would be cool.
Too bad you don't have a q6600 for comparison.

Can get those clocks but can't change NB/HT to high anymore. OD and BIOS problems. Max they'll be is in-sync 1:1 at around 2.1GHz.

Great, how do you measure the card?

Did it on an older system which has onboard around 3 weeks ago. Connect one peripheral at a time, disconnect one peripheral at a time and so on whilst knowing the chipset/RAM/HD power draw.

Looking at your screenshot it seems CnQ does not reduce the HT multiplier.
You said HT speed has an big impact on consumption, so with a lower HT multiplier idle power should drop a little.
So i'm still confident a 9500 system will require less power than an q6600 in idle once CnQ (2.0 as in thw's review) works flawless.

Yep, it keeps NB/HT speed the same. Volts drop 0.2V on CPU, and multi drops, that's it.

Q6600 should be higher than X4 9500 IMO. I did tests on C2D before:

E6750 65W TDP (G0)
1.350V 2.67GHz stock
HD 2600 XT (stock)
2x1GB Corsair XMS2 C4 PC2-6400 @ 2.1V
WD Caviar SE 80GB Sata II
3 readings for each over a 10 minute period

Power Idle: 93-94-95W
Power CPU0 load w/Prime95: 137-137-137W
Power CPU0+1 load w/Prime95: 158-160-161W

This X4 I have now is RAM running at 2.2V and one more WD Caviar SE SATA II HD.

I can't use OD so dropping HT/NB is going to be problematic now since if I do it in the BIOS I have to reset all values again.

PS. 9500 finally on the road, should be here in a few hours. I'm really curious how far i will get in the M2A-VM, max. ref HT i could reach with an X2 3800+ EE was 273MHz so i'm confident to reach the cpu limit.

I reckon you should be able to reach around that much on the 9500. Good luck.

Here's some more bandwidth/latencies for ya (note I can only boot maximum 1:2 divider): http://valid.x86-secret.com/show_oc.php?id=276891

[saaya]

justapost,
max ht you could reach was 273 with a 3800+?
did you try other ht multipliers and cpu multipliers?
273 is really really low for a k8/k9 cpu...

[Andi64]

I've worked around it and here are my findings:
I found out the issue is NOT the NB or CPU speed BUT the HT ref. speed.
The maximum for everyone I've seen is around 250-268 HT ref. Mine is 267MHz HT. ref.

I'm having the same results with K8. HT @ 250Mhz it's very easy with stock volts, 255Mhz it's completely unstable...
I think it's the chipset... do you have a K8 to test the motherboards maximun HT?

Thanks!

[justapost]

justapost,
max ht you could reach was 273 with a 3800+?
did you try other ht multipliers and cpu multipliers?
273 is really really low for a k8/k9 cpu...

Actualy the board can go up to 282 with a lower multiplier, 270-290 is max ref HT reported for M2A-VM.

EDIT: With 9500 max i can reach in M2A-VM with stock cooler is 235 ref HT but i need CPU 1,5V NB 1,4V and Ram 2,1V for that and the cpu temp is around 83° after ten minutes prime95 (cpu-z id 277066).
HT 1175 MHz, NB 2115 MHz, Mem 470 MHz ganged 5-5-5-15-21 2T.
Will run benchmarks now, but i think it's not worth a new tread till i get a better cooler.
I need 1,45V for 230 and 1,375 for 220, at 1,25 V max i can reach is 205..
Will grab an volt meter tomorrow and must wait for better cooler and M3A board now.

[KTE]

Hah just found out how many pics I had posted

Anyway, all is rehosted now, thanks.

I'm having the same results with K8. HT @ 250Mhz it's very easy with stock volts, 255Mhz it's completely unstable...
I think it's the chipset... do you have a K8 to test the motherboards maximun HT?

Yep, but I've not oc'd it yet. I want to try and get a better BIOS before I pop that in, but I'll probably use 113 BETA and try the X2 tomorrow.

Hopefully board can do +300MHz HT.

[KTE]

Warning: Verbose post ahead.

I wanted to quickly explain AM2+ - 790FX overclocking.

NOTE: The below method worked for BIOS v1.1, P0C, P0D and 113 BETA. Putting an NB multiplier higher than 9x made the system start downclocking in 0.5x multiplers the CPU and HT frequency.

I told a while ago of the HT ref. limit of ~265-275MHz max on Phenom 9500 so far which limits us to a maximum of 2.9-3.0GHz screenable and still very very unstable. I've been messing about with AM2+ P-States for 2-3 days now. I haven't seen others do it yet here and I got the feeling very few actually understand it, but I did post a 3GHz NB with 2.1GHz HT link speed 2 dys ago and no one really caught on to how or what was happening.

I wanted to explain it for those that are unsure but I needed some more testing time and different BIOS to be able to experiment more and see if it makes a difference. Then the last of my queries was answered my Macci in a thread yesterday (CPU VID): make sure you read it!

Here is how AM2+ P-States work.

790FX BIOS give you extra options to control:

CPU FID
CPU VID
CPU DID

NB FID
NB DID
NB VID

=Frequency ID
=Voltage ID
=Divider ID

Plus you can control:

HT speed multi
HT reference speed

What FID does is decide your clock frequency multiplier.

So for instance, CPU FID 02 equals 9x CPU multiplier (03 is 9.5x, 04 is 10x and so on).

BUT changing FID alone will not be equal to changing the CPU multiplier. There is still part of the equation missing.

There is CPU DID which is the Divider for the FID value.

So, you have options of 1, 2, 4, 8 and so on for CPU DID. What they do is divide the end CPU FID value, and then the HT reference speed is multiplied by it to gain the CPU speed.

Hence, if we input CPU FID value 08 which stands for CPU multiplier 12x, and we choose CPU DID value 1 (which is what its says), then:

FID / DID = CPU Multiplier
12 / 1 = 12x multiplier

If we chose CPU DID 2 then the end CPU multiplier would've been 6x.

So the HT reference clock speed you set (say 200MHz as an example) will multiply x12 to get 2400MHz CPU speed now.

HOWEVER the 9500 I have is multipler locked obviously. So whatever FID/VID value you choose above what gets you 11x multiplier will just result in no change from stock 11x multipler.

CPU VID is the Voltage identity which I did't fully understand till Macci replied and said to leave it at default which is value 28 for 9500 and stands for 1.250V. If you increase it to value 29 you get 1.1875V with my testing.

BUT if you increase the voltage (i.e. decrease the value to 27), you will fail POST.

Now we know how CPU speed is worked out.

Additionally, NB FID, DID, VID are of the same task but for the NB instead and NB VID can be changed to higher. Value 36 is stock which for me was 1.1V.

NB FID values are as so: 00 is 5x NB speed multiplier, 01 is 6x, 02 is 7x and so on.

For NB DID you get two options of divider 1 and divider 2.

What that NB FID does is, get the HT reference clock you set (say 200MHz) and multiply that by the NB multiplier you choose through NB FID / NB DID.

So say you choose NB FID of value 00 which is 5x multi and choose NB DID of value 1 (which is again what it says).

NB FID / NB DID = NB Multiplier
In the above case we have 5 / 1 = 5x multiplier

With a 200MHz HT ref. speed, that would give us 200x5 = 1000MHz NB speed.

BUT=> NB clock speed can only be equal or higher than HT speed. Keep that in mind.

If it isn't, then either you'll have no boot or the system will downclock itself lower than stock. If you set a high NB clock, your system may not boot at all or downclock the NB/CPU/HT to make it bootable.

Now we know how the NB speed is worked out.

Easy as that! Any qs, please ask. I've tried to make it simple but you'll have to give it a go before understanding properly.

Here was my older run using this method I posted within those images

[informal]

Nice work man!IMO,this should be stickied ASAP

[nguoisaigon]

thanks for your information KTE. But I have just returned the Phenom 9600 back to AMD distribution in VN. I'll get it back a few days later (around 6 or 7 this month). Then I'll test it more. I have seen a CPUZ validation of a 6400+ Black Edtion @ 520 HTT Ref with K9A2 Platinum.

[2good4you]

Very interesting to see how the NB (IMC/L3) can be clocked totally independently from the cpu itself. Phenom should be fast if someone could run all at sync if so is possible, let's say cpu 3GHz and also the NB at 3GHz. Anyway it's an interesting architecture, and nice posts KTE, keep the good work up! One question, can you in any 790FX motherboard set the NB speed separate from cpu in the bios and not with apps in win like WPCREDIT?

[JumpingJack]

Nice work man!IMO,this should be stickied ASAP

Agreed. This is just too much good info.

[KTE]

Thanks all. I thought it's useful to add in there because I understand some users will be shy to ask.
As I posted above (12-02-2007 11:43 AM) the difference in L2/L3 cache latency by increasing NB speed is quite a lot. A drop of more than half the cycles is possible.

I've tested more: my CPU is bugged out and slower than others here after around 2.45-2.55GHz. Even at faster NB/HT and RAM clocks than them, it performs worse. I kinda knew this was coming all along and have been testing it for a while now. I think I'm going to get rid of this X4 now.

Also, the MSI K9A2 v1.1 BIOS doesn't change P-State values properly. 113 Beta (aka 1.1B3 IIRC) works.

One question, can you in any 790FX motherboard set the NB speed separate from cpu in the bios and not with apps in win like WPCREDIT?

Yes, that's how I'm doing it, through the BIOS.

I'll post a picture of of the P-State edit menu soon. It's something which has been added very recently and is still quite buggy.

[KTE]

Look HERE for pics of BIOS options, AM2+ P-States and BIOS Overclocking screenshots.