AMD FX-9370 Vishera 4.4GHz

[RGone]

Yes we try to stay away from as much crap as possible. A lot of that comes from the people that help being accustomed to their own rigs quite well and talking amongst ourselves outside the forum. This tends to give off a more realistic picture of our computer world. Less speculation.

There were some 6 or 7 of us that knew what our rigs would do and we started to compile the P-states to see if there were a pattern in that maybe a lower P-state voltage at 4.0Ghz did in fact make for a better overclocker. I was asked to rank the 6 or 7 that had given their P-states and being familiar with their systems and how they actually ran them day in and day out...I was able to do so pretty accurately.

To try and do so without full knowledge of how any other systems were being run daily, it would not be so easy to 'rank' other cpus. Need full on information about the system and how fast it is run day in and day out.

We believed that a lower voltage for the 4.0Ghz P-state would mean a better clocker but from just sorting the 6 or 7 known rigs that we had good information on; that theory was not proven true more than maybe 70% of the time. I suggested that such a theory would only be useful when buying a used cpu anyway since we had no knowledge of the P-states of new in the box cpus anyway. I also said that if I had two cpus side by side and knew the P-states, of each, my gut would say go for the cpu with the lower voltage for the 4.0Ghz state.
RGone...ster.

Does this mean a lower voltage reading on the 20.00x line would indicate a good chip and a possible higher stable clock?

P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0D - IDD 13 (20.50x - 1.387 V)
P-State FID 0x18 - VID 0x13 - IDD 12 (20.00x - 1.313 V)
P-State FID 0x12 - VID 0x1B - IDD 10 (17.00x - 1.212 V)
P-State FID 0xC - VID 0x23 - IDD 8 (14.00x - 1.112 V)
P-State FID 0x5 - VID 0x2C - IDD 5 (10.50x - 1.000 V)
P-State FID 0x10C - VID 0x36 - IDD 4 (7.00x - 0.875 V)


I joined this forum back in May and my mistake not to do some reading here. I feel like I have learned more in the past two days than the previous 8 months. Particularly in regards to the GA990FXA-UD3 Rev3.0 and it's shortcomings. It's been a pleasure to read an adult conversation about this. Other places you need to filter out a lot of crap.

 

[Johan45]

Does this mean a lower voltage reading on the 20.00x line would indicate a good chip and a possible higher stable clock?

P-State FID 0x1A - VID 0x0A - IDD 13 (21.00x - 1.425 V)
P-State FID 0x19 - VID 0x0D - IDD 13 (20.50x - 1.387 V)
P-State FID 0x18 - VID 0x13 - IDD 12 (20.00x - 1.313 V)
P-State FID 0x12 - VID 0x1B - IDD 10 (17.00x - 1.212 V)
P-State FID 0xC - VID 0x23 - IDD 8 (14.00x - 1.112 V)
P-State FID 0x5 - VID 0x2C - IDD 5 (10.50x - 1.000 V)
P-State FID 0x10C - VID 0x36 - IDD 4 (7.00x - 0.875 V)


I joined this forum back in May and my mistake not to do some reading here. I feel like I have learned more in the past two days than the previous 8 months. Particularly in regards to the GA990FXA-UD3 Rev3.0 and it's shortcomings. It's been a pleasure to read an adult conversation about this. Other places you need to filter out a lot of crap.

I'm still curious as to what clocks and voltage you run on the 8350 daily? Even if we can't get a huge list going without more details, It would be interesting. Since I think that is one of the lowest we have seen yet.

 

[RGone]

I'm still curious as to what clocks and voltage you run on the 8350 daily? Even if we can't get a huge list going without more details, It would be interesting. Since I think that is one of the lowest we have seen yet.

For sure plus we need to know what P-state links to what cpu. He has FX-8350 and an FX-9590 but did not specify which his listed P-state was referring to.
Bobert...ster.

 

[marjamar]

Here's the P-States on this FX9370. Don't know if they're good or bad, but I am running a stable 5.097 since last night without a hitch. Haven't really done much to test it yet, but I did run prime95 for about 20 minutes or so while watching temps and it kept humming along. I'm on an EK waterblock with 360 rad on 6 push-pull fans, so CPU temps maxed at about 62c and core at about 55c. I'll be testing more over the next few days and record some of the screens and bios settings.

Anyways, here the P-States -- All stock settings in BIOS.

P-State FID 0x1F - VID 0x01 - IDD 18 (23.50x - 1.537 V)
P-State FID 0x1C - VID 0x01 - IDD 18 (22.00x - 1.537 V)
P-State FID 0x1A - VID 0x01 - IDD 18 (21.00x - 1.537 V)
P-State FID 0x16 - VID 0x0A - IDD 14 (19.00x - 1.425 V)
P-State FID 0x12 - VID 0x14 - IDD 12 (17.00x - 1.300 V)
P-State FID 0x8 - VID 0x24 - IDD 7 (12.00x - 1.100 V)
P-State FID 0x10C - VID 0x35 - IDD 4 (7.00x - 0.887 V)

-Rodger

 

[Mandrake4565]

We believed that a lower voltage for the 4.0Ghz P-state would mean a better clocker but from just sorting the 6 or 7 known rigs that we had good information on; that theory was not proven true more than maybe 70% of the time. I suggested that such a theory would only be useful when buying a used cpu anyway since we had no knowledge of the P-states of new in the box cpus anyway. I also said that if I had two cpus side by side and knew the P-states, of each, my gut would say go for the cpu with the lower voltage for the 4.0Ghz state.
RGone...ster.

Another thing to add to the Gonesters comment is that only a few of us have really pushed these things as far as we can realistically push them. So if looking to run high stable overclocks, it seems that with adequate cooling, a low voltage P-State chip will run a "high" 24/7 overclock with less volts, then one with a higher P-State. To the point though, from what I've seen, so far the people with the higher P-State chips seems to be able to overclock them higher, as far a max clocks non stable goes. An example I can refer to is Johans and My 8350's, his has reached higher max clocks then mine. I've done 5475 mhz superpi runs on mine where as he has been able to do 5668 mhz. Though this may also be because of better cooling.

 

[madcratebuilder]

I'm still curious as to what clocks and voltage you run on the 8350 daily? Even if we can't get a huge list going without more details, It would be interesting. Since I think that is one of the lowest we have seen yet.

Sorry I didn't specify what cpu. Those p-states are from the 8350. I normally run at 4400 with 24/7 stability. The h80i keeps the temps 32-38c with under 1k fan rpm.

I have two oc's stored, one is 4515, runs fine for everyday 'puting, temps are great but it drops core #6 and core #8 when I run P95. I have to push fan speed to 2k+ rpm when stress teasting.

I have a 4664 oc that runs similar but well drop 3 and eventually 4 cores with p95,

All these are with default voltages. To tell the truth I"ve been hesitant to change voltage as I'm not sure what I'm doing. I've been digesting threads here and feel more comfortable about upping the voltage.

I can control core temps with rad fan rpm, if I push both fans to max, about 2550 rpm, temps are never over 55c when stress testing.

I've been reading about the problems with the GA990fxa-ud3 Rev3.0 boards, all this time I thought I was just not figuring it out. It appears my Rev3.0 board is not one of the better ones.

 

[mtdewcowboy]

Another thing to add to the Gonesters comment is that only a few of us have really pushed these things as far as we can realistically push them. So if looking to run high stable overclocks, it seems that with adequate cooling, a low voltage P-State chip will run a "high" 24/7 overclock with less volts, then one with a higher P-State. To the point though, from what I've seen, so far the people with the higher P-State chips seems to be able to overclock them higher, as far a max clocks non stable goes. An example I can refer to is Johans and My 8350's, his has reached higher max clocks then mine. I've done 5475 mhz superpi runs on mine where as he has been able to do 5668 mhz. Though this may also be because of better cooling.

I will let ya know how high I get once my WC setup is in

 

[Johan45]

Sorry I didn't specify what cpu. Those p-states are from the 8350. I normally run at 4400 with 24/7 stability. The h80i keeps the temps 32-38c with under 1k fan rpm.

I have two oc's stored, one is 4515, runs fine for everyday 'puting, temps are great but it drops core #6 and core #8 when I run P95. I have to push fan speed to 2k+ rpm when stress teasting.

I have a 4664 oc that runs similar but well drop 3 and eventually 4 cores with p95,

All these are with default voltages. To tell the truth I"ve been hesitant to change voltage as I'm not sure what I'm doing. I've been digesting threads here and feel more comfortable about upping the voltage.

I can control core temps with rad fan rpm, if I push both fans to max, about 2550 rpm, temps are never over 55c when stress testing.

I've been reading about the problems with the GA990fxa-ud3 Rev3.0 boards, all this time I thought I was just not figuring it out. It appears my Rev3.0 board is not one of the better ones.

You do need to set the volts manually, I would suggest you try setting a voltage of 1.4V_core and see how high it can take you stable. If P95 is dropping cores then you're not stable.

Here's the P-States on this FX9370. Don't know if they're good or bad, but I am running a stable 5.097 since last night without a hitch. Haven't really done much to test it yet, but I did run prime95 for about 20 minutes or so while watching temps and it kept humming along. I'm on an EK waterblock with 360 rad on 6 push-pull fans, so CPU temps maxed at about 62c and core at about 55c. I'll be testing more over the next few days and record some of the screens and bios settings.

Anyways, here the P-States -- All stock settings in BIOS.

P-State FID 0x1F - VID 0x01 - IDD 18 (23.50x - 1.537 V)
P-State FID 0x1C - VID 0x01 - IDD 18 (22.00x - 1.537 V)
P-State FID 0x1A - VID 0x01 - IDD 18 (21.00x - 1.537 V)
P-State FID 0x16 - VID 0x0A - IDD 14 (19.00x - 1.425 V)
P-State FID 0x12 - VID 0x14 - IDD 12 (17.00x - 1.300 V)
P-State FID 0x8 - VID 0x24 - IDD 7 (12.00x - 1.100 V)
P-State FID 0x10C - VID 0x35 - IDD 4 (7.00x - 0.887 V)

-Rodger

Wow I wonder if this tren will carry over to the 9xxx series as well. I assumed that since they were "cherry" picking the silicon, they would be very similar but comparing marjamar's to mine I would absolutely say that isn't so. And so far from what I have seen I do have a "good" FX-9370. Here's mine for an easy comparison.

# of P-States 7
P-State FID 0x1F - VID 0x04 - IDD 18 (23.50x - 1.500 V)
P-State FID 0x1C - VID 0x04 - IDD 18 (22.00x - 1.500 V)
P-State FID 0x1A - VID 0x04 - IDD 18 (21.00x - 1.500 V)
P-State FID 0x16 - VID 0x0D - IDD 14 (19.00x - 1.387 V)
P-State FID 0x12 - VID 0x16 - IDD 12 (17.00x - 1.275 V)
P-State FID 0x8 - VID 0x26 - IDD 7 (12.00x - 1.075 V)
P-State FID 0x10C - VID 0x36 - IDD 4 (7.00x - 0.875 V)

 

[madcratebuilder]

You do need to set the volts manually, I would suggest you try setting a voltage of 1.4V_core and see how high it can take you stable. If P95 is dropping cores then you're not stable.

Following your advise I set vcore to 1.440. I now can run P95, all 8 cores stable at 4.62ghz. The 80i works hard to keep the temp just under 60C.

What is the max safe temp for the 8350? I have read a few different statements on this. I have been working with 62C for cores and 70C for socket. I'm I even close?

I am seeing some short, random throttling, artifact of this GA990fxa I believe. I have all power saving off in bios.

How high can one go on the vcore voltage?

I'm enthused about OC'ing again, thanks.

 

[RGone]

The 80i works hard to keep the temp just under 60C. = BUT which of these two temps are you calling 60C? I have been working with 62C for cores and 70C for socket. I'm I even close? << That is prefectly well what we all use.

Is HPC menu option available in that bios. It might help in that throttling deal. Enable HPC.
RGone...

 

[Mandrake4565]

How high can one go on the vcore voltage?

It all depends on what your intentions are, AMD says it's 1.5. You'll probably run out of temp head room before you come close though.

 

[marjamar]

This CPU is not stable at anything less then 1.5 volts it seems unless I'm below 4.7GHz. Stock voltage is around 1.54 or something. I believe that is above the recommended to volts for this chip. RMAing it back for a refund to newegg. I'd not recommend getting these overclocked chips, as you simply do not know how much they overvolted them to get their clocks stable.

Had to fight with newegg for a refund too. Must be alot of returns on this thing as I never had to fight with newegg before. Good thing I've spent as much as I have with them over the 10 years, or I may have not been issued a refund.

-Rodger

 

[bassnut]

If my memory is right ...... When Johan got his FX9370 he had to work hard at getting stable above 4.8GHZ regardless what he set for VCore. In the end the problem turned out to be with the ram timings causing his issues. I am sure he can shed more light on this then I can. If you have the time Check out the Just "Putzin Around Thread" where Johan posted his OC on the chip.

http://www.overclockers.com/forums/showthread.php?t=731320&page=27

 

[Johan45]

You were close Bassnut it was 5.0 I was having trouble with. I just didn't want to keep pumping the volts to it.
@ marjamar, I did notice that in your P-states it was listed for 1.537. so yes that is default voltage for your chip. When I started with mine I set the voltage to 1.45 then saw how hi I could go starting with 4.5G. Too bad you didn't give it more of a chance.

 

[Hr.Soegaard]

Nice to follow this thread. I just maybe have a stupid question.
How to tell if your overclocked 8350 is throttling?
I love overclocking and have learn a lot from u guys, but after reading this:

"I am seeing some short, random throttling, artifact of this GA990fxa I believe. I have all power saving off in bios."

I was thinking, should that be for my gfx or not the cpu.?

Thx.

 

[10XTriplet]

There are plenty of review's of the 8-core FX-9590 @ 4.7Ghz vs 4-core 4770K @ 3.5Ghz. Not to make the an AMD vs Intel thread, but that's what most people do is compare... However, the FX-9590 is 220 watts without overclocking it.. Daaaamn!

I'm impressed that it keeps up pretty good...

 

[Johan45]

Nice to follow this thread. I just maybe have a stupid question.
How to tell if your overclocked 8350 is throttling?
I love overclocking and have learn a lot from u guys, but after reading this:

"I am seeing some short, random throttling, artifact of this GA990fxa I believe. I have all power saving off in bios."

I was thinking, should that be for my gfx or not the cpu.?

Thx.

Are you asking or advising??
CPU throttling will be noticeable if your pushing it and monitoring the CPU speed. The speed will fluctuate up and down.

 

[Hr.Soegaard]

ahh okay, so under prime95 i will see it jump up and down.

 

[madcratebuilder]

Is HPC menu option available in that bios. It might help in that throttling deal. Enable HPC.

I Think I disabled HPC, I was thinking it may have been causing the occasional throttling I'm seeing. I'll go back and enable.

This 4.62 is all the 80i can handle and keep the core under 60C. As soon as I clean up one or two projects here the 990fxa/8350 is moving to another case, I had planned on keeping the 80i with it but I have a 100i setting here unopened that I may use. I think the 100i can remove considerably more heat. That well require some major case work.

I think for now I well work on my memory, I have Corsair 1866 and it well only run at 1400 with the 990fxa.



ETA:

How to tell if your overclocked 8350 is throttling?

Watching the individual cores and reported speeds in HWInfo64. I see it drop to about 1.4 then right back up to 4.6. It may be related to core temps, seems to do it more often as the core approaches 60C.

 

[marjamar]

I would have put more time in it, but I could see it was not going to run with less then 1.5 at any clock I would be interested in. I'm back on the FX8150 now and it is running at 4.773 GHz (my default 24/7) and core voltage is 1.488. The FX-9370 would not run this without hanging on anything more then minor loads. Also, I am on a very nice water loop and block, so heat really wouldn't have been an issue I don't believe.

As to memory, I do run it at lower clocks normally at the start of testing a CPU. I get a feel for the multiplier/volts and then mix in the memory clocks/timings along while bumping buss up and multiplier down. The FX-3070 seemed fairly stable at 5.076, but had voltage on auto to see where it went and most of the time it was up about 1.54 and even see it bump up against 1.6 under prime95. That is simply too high a voltage, even on a good water loop. I was up to 70c on the socket and hit 60c a couple times on a few of the cores. That's too much heat. So what was left. Eat it or send it back was all I could think.

If my memory is right ...... When Johan got his FX9370 he had to work hard at getting stable above 4.8GHZ regardless what he set for VCore. In the end the problem turned out to be with the ram timings causing his issues. I am sure he can shed more light on this then I can. If you have the time Check out the Just "Putzin Around Thread" where Johan posted his OC on the chip.

http://www.overclockers.com/forums/showthread.php?t=731320&page=27