Overclock.net banner

FX-9590 show and tell

11K views 118 replies 23 participants last post by  Kryton 
#1 ·
Hello fellas.

Here is some eye candies for you. Some of you've seen this, else where, but thought it would be a good addition to the forum.

In search of lower temps.

For honesty purposes, the FX-9590 is hot. I don't mean it's warm or just hot, I mean hot!

I toyed with this cpu for some time. Learning the board and the chips abilities.

Much time has been spent with an automatic everyday user purchase and use scenario. (pretty much for me to game with
biggrin.gif
) and also to enjoy a guarantee clock speed of 4.7ghz and a multi core turbo boost of 5ghz.

Now I pulled the P-state list..... here it is

# of P-States 7
P-State FID 0x22 - VID 0x02 - IDD 18 (25.00x - 1.525 V)
P-State FID 0x1F - VID 0x02 - IDD 18 (23.50x - 1.525 V)
P-State FID 0x1D - VID 0x02 - IDD 18 (22.50x - 1.525 V)
P-State FID 0x18 - VID 0x0B - IDD 14 (20.00x - 1.412 V)
P-State FID 0x12 - VID 0x15 - IDD 11 (17.00x - 1.287 V)
P-State FID 0x8 - VID 0x25 - IDD 7 (12.00x - 1.087 V)
P-State FID 0x10C - VID 0x35 - IDD 4 (7.00x - 0.887 V)

WUT??

talking 1.52500v at only 22.5X multi.

Want truth?

This processor doesn't remain stable under manual overclock conditions and trying to under volt a Cpu VID of 1.52500v

It causes freezes or system hangs like a SSD or HDD is going bad..... It's actually just a low volt typical AMD hang. No errors in testing, no overheats.... just hangs.

To make a long story shorter as I'm pressed for time this evening.......

At stock 4.7ghz with turbo and all green enabled, truth be told this processor would heat soak my 120.2 X2 rad set up cooling only the Cpu. Also in recognition the water loop is old. Rads have many years on them. Usually only run 5 rad fans as I am still doing. So essentially I was seeing thermal throttling between 30-45 minutes of stress testing.

65c Cpu socket temp - throttles. 90c Core temp, board shuts completely off.

Cpu would drop to 4.5ghz. I never say this gaming.... my games seldom use more than 2 cores. I don't have SLI, and I play mostly DX-9 games, One of my favorites being Hawken to utilize some awesome physx. Running a GTX 760, frame rates are always good even full physx.

Any how.... I wanted to eliminate that 65c thermal throttle with the stress testing but in reality it never happens just browsing and posting pictures like I will be doing very shortly here.

Any how, the de-lid was a success. Generally speaking Cpu socket temps stay 45c. With the stress testing I can keep her under 60c and this IS dependent on ambient temps. Nice warm loop, needs a rebuild.... but I saved my lazy butt some time on that fun!

Heres pictures, would be very happy to answer questions. Please keep in mind this is a daily rig, I seldom overclock it (doesn't need it) and enjoy very much the better temps!!







 
See less See more
8
#2 ·
that's interesting. I thought all amd chips were soldered?
 
#4 ·
Nice work !

Some things got me surprised when comparing my O.C with your little monster. My trusty FX-8350 "only" need 1.464v~1.476v to be perfectly stable at 4.9 Ghz. All cores. Your cpu load vCORE uses 1.525v ?
devil.gif


Too much voltage to me.

Something tells me that your chip can undervolt a little bit.

Other, my socket temps throttling are 72ºc+. I never got bogged with throttling under this temp.

Perhaps a motherboard behaviour? I do have extra cooling in the N.B and Socket (Fan on the Socket behind the motherboard)

Anyway great chip and RIG ! Grats !
 
#5 ·
Quote:
Originally Posted by jclafi View Post

Nice work !

Some things got me surprised when comparing my O.C with your little monster. My trusty FX-8350 "only" need 1.464v~1.476v to be perfectly stable at 4.9 Ghz. All cores. Your cpu load vCORE uses 1.525v ?
devil.gif


Too much voltage to me.

Something tells me that your chip can undervolt a little bit.

Other, my socket temps throttling are 72ºc+. I never got bogged with throttling under this temp.

Perhaps a motherboard behaviour? I do have extra cooling in the N.B and Socket (Fan on the Socket behind the motherboard)

Anyway great chip and RIG ! Grats !
It a 9590 amd put a 1 years warantly on the cpu and it max voltage from factory is about 1.525v of what i know to be stable.

And 4.9 and 5.0 need completly different voltage anyways
 
#6 ·
All FX chips are soldered.

I've De-lidded the following soldered processors successfully with ok success. Couple of duds on the list here.

FX-6100 De-Lidded Working
FX-6300 De-lidded Working (contest prize giveaway)
FX-9590 De-lidded Killed x1
FX-9590 De-lidded Working (second attempt)
FX-8320 De-lidded Sold Working
Athon 5000+ unlocker FX-5000 Kept
970T De-lidded for a customer Working
1090T De-lidded Sold Working
965BE De-lidded Sold working
940BE De-lidded Sold working
9850BE De-lidded Sold working
9950BE De-lidded Sold working
FX-4100 De-lidded NOT Working
6400+ De-lidded Sold working
6400+ delidded working kept (2nd 6400+ delidded)

______________________________________________________________________

A mention on surface area and heat accumulation:

I found through many of these processors, the IHS plate was an accumulator more so then a dissipater'. Increasing surface area works pretty good for most air coolers with 95w chips and greater. Helps less for water cooling lads. The reason being the water block is already an accumulator of heat, the water is the dissipater. The heat transfer through solder is good, better than thermal paste, but when using solder, IHS plate then some TIM, you begin to slow the heat dissipation process thus having higher socket temps. I found bringing cooling directly to the source of heat to be outstanding performance. Dropping max temps enough to fully gain on average 100mhz or higher Cpu speed. In addition to this, I've also found de-lidding to bring Core and Socket temps much closer together and seemingly better stability while loading and general usage plus gaming.

Quotable answers:
My trusty FX-8350 "only" need 1.464v~1.476v to be perfectly stable at 4.9 Ghz. All cores. Your cpu load vCORE uses 1.525v ?

Technically yes. I can however load test 1.49500v at 4.9ghz and it's stable until "eventual" low volt hang happens. Processor doesn't receive requested (needed) pstate voltage at xclock at certain time interval, it locks right up. No errors, no blue screens... just a simple hang. Not the end of the world, I've spent hours gaming, generally 2-3 hours and nothing happened, walk away for a few hours to come to a hung PC.

While running this chip on the bench with some chilling involved, 5600mhz benchmarking and gaming was pretty easy while maintaining much lower temps.

Also have done tweaking with sub 200 reference clock speeds and just had a joy of a time. It's a fun overclock with FX without doubt there's no more to be said with AMD FX processors.

So here's a picture 4875mhz 1.488v and some Cinebench running on the daily loop. This configuration also was likely to hang with some point of time, it's more a when then why. That dang VID is high up there.

I think Stilt has answered many questions on leaky transistors. All I know, this chip loves voltage and doesn't seem to leak as much as require it.

 
#7 ·
Quote:
Originally Posted by ShrimpBrime View Post

All FX chips are soldered.

I've De-lidded the following soldered processors successfully with ok success. Couple of duds on the list here.

FX-6100 De-Lidded Working
FX-6300 De-lidded Working (contest prize giveaway)
FX-9590 De-lidded Killed x1
FX-9590 De-lidded Working (second attempt)
FX-8320 De-lidded Sold Working
Athon 5000+ unlocker FX-5000 Kept
970T De-lidded for a customer Working
1090T De-lidded Sold Working
965BE De-lidded Sold working
940BE De-lidded Sold working
9850BE De-lidded Sold working
9950BE De-lidded Sold working
FX-4100 De-lidded NOT Working
6400+ De-lidded Sold working
6400+ delidded working kept (2nd 6400+ delidded)

______________________________________________________________________

A mention on surface area and heat accumulation:

I found through many of these processors, the IHS plate was an accumulator more so then a dissipater'. Increasing surface area works pretty good for most air coolers with 95w chips and greater. Helps less for water cooling lads. The reason being the water block is already an accumulator of heat, the water is the dissipater. The heat transfer through solder is good, better than thermal paste, but when using solder, IHS plate then some TIM, you begin to slow the heat dissipation process thus having higher socket temps. I found bringing cooling directly to the source of heat to be outstanding performance. Dropping max temps enough to fully gain on average 100mhz or higher Cpu speed. In addition to this, I've also found de-lidding to bring Core and Socket temps much closer together and seemingly better stability while loading and general usage plus gaming.

Quotable answers:
My trusty FX-8350 "only" need 1.464v~1.476v to be perfectly stable at 4.9 Ghz. All cores. Your cpu load vCORE uses 1.525v ?

Technically yes. I can however load test 1.49500v at 4.9ghz and it's stable until "eventual" low volt hang happens. Processor doesn't receive requested (needed) pstate voltage at xclock at certain time interval, it locks right up. No errors, no blue screens... just a simple hang. Not the end of the world, I've spent hours gaming, generally 2-3 hours and nothing happened, walk away for a few hours to come to a hung PC.

While running this chip on the bench with some chilling involved, 5600mhz benchmarking and gaming was pretty easy while maintaining much lower temps.

Also have done tweaking with sub 200 reference clock speeds and just had a joy of a time. It's a fun overclock with FX without doubt there's no more to be said with AMD FX processors.

So here's a picture 4875mhz 1.488v and some Cinebench running on the daily loop. This configuration also was likely to hang with some point of time, it's more a when then why. That dang VID is high up there.

I think Stilt has answered many questions on leaky transistors. All I know, this chip loves voltage and doesn't seem to leak as much as require it.

What mean contest price giveaways? did you win it?

Should try a fx 8300 they are cheap
smile.gif


and yea ihs is worth nothing that why gpu dont have them
 
#9 ·
Quote:
Originally Posted by ShrimpBrime View Post

Yes FX-8300. Good chip. LN2 on sub 50$ motherboard.

http://valid.canardpc.com/cih7iw

I gave the processor away at a contest held at my HWBot team site ClassicPlatforms, IMOG was the winner of FX-6300


Awesome
biggrin.gif


Poor mobo is the vrm are with ln2 too never really looked to know if they where on motherboard in o/c record I know gpu are.
 
#15 ·
Quote:
Originally Posted by ShrimpBrime View Post

You are young man to still have grand parents. Mine have been gone for a long time.

You'd be better to build de-lidded i3 eh?
What we can delid i3 oh well could be fun but i will have my h75 to put to some use and the 8300 was a good choice in my book ((( im going to drop some gpu in the build for folding@home So the extra multithreaded perf would be nice

But yea a i3 would be cheaper
biggrin.gif
 
#16 ·
Nice run Shrimpy. We have the same motherboard and CPU and I set everything to Auto in my BIOS except for manual memory frequency and timings because if I touch anything else, my system hangs. And yes, 9590 is like trying to cool down lava. I usually get around that 724 score you see in orange, which is running at everything stock, 4716MHz.

Here:
 
#19 ·
Quote:
Originally Posted by MrElusive View Post

Nice run Shrimpy. We have the same motherboard and CPU and I set everything to Auto in my BIOS except for manual memory frequency and timings because if I touch anything else, my system hangs. And yes, 9590 is like trying to cool down lava. I usually get around that 724 score you see in orange, which is running at everything stock, 4716MHz.
I had done a whole bunch of testing about this "system hang" issue. I mean I've loaded all kinds of overclocks, underclocks DOCP settings, loaded PC chip settings, up to 275 bus clocks, under 200 bus clocks...

Even had hard times with FID 0x18 - VID 0x0B - IDD 14 (20.00x - 1.412 V) and trying to under volt this here..... No good. Must be running 1.41200v or higher at 20x.

Then you look at VID for 22.5x and higher... requests 1.52500v-

If you can cool that 1.5250v VID, you'd be able to run all 8 cores at 5ghz all day long. But them temperatures......

Honestly this set up is for massive clocks with LN2. I'd be willing to bet most FX-9590 could hit 8ghz with CHVZ motherboard.

Quote:
Originally Posted by bonami2 View Post

It me or a 9590 would be perfectly cooled with a h110i gt

im going to drop some gpu in the build for folding@home So the extra multithreaded perf would be nice
Hopefully you make grandmother aware her PC is going to be folding and pulling 600w at the outlet and her electric bill is going to increase.

Also is violation of EULA.

Re-distributing the FAH files or packaging the FAH files inside another software package in a attempt to install FAH with or without the user's consent is a violation of the EULA. http://folding.stanford.edu/home/faq/faq-policies

This comment H110i cool a FX-9590 for your 2000th post??

Umm No.
 
#22 ·
Quote:
Originally Posted by FlawleZ View Post

Yeah I considered a new build with FX-9590 but with only H100i GTX cooling would be lacking. How much worse is the stock cooler?
It doesn't get shipped with enough cooling. I believe it's a 120.1 rad.... complete garbage. AMD should have stuck with selling this chip without a cooler and suggested for Custom water loops only.

With the Corsair H100 and the likes of these coolers with set up defaults, the cpu will just simply throttle. (during load testing in most cases)

additional comment:

Have been reading many issues with heat with FX-9590. Most people with AIOs seem to have this issue mostly. Cost to performance is good, but not enough for 220w and beyond dissipation.
Quote:
Originally Posted by Casey Ryback View Post

That's eye candy for sure. Great job.

Not impressed by the 9590 that is just an overpriced 8300 no?
Thanks-
Overclocked FX-8300? That's an interesting way to look at it....

FX-8300 failed any testing to become a stable FX-8320 during binning. There's a reason it's clocked 3.3ghz at 1.1v.

FX-9590 passed 4.7ghz binning....
 
#24 ·
Quote:
Originally Posted by CS14 View Post

Very nice thread. Making me consider delidding my 8320 at some point in the future when I eventually go custom water cooling and really give it a good push. Should make a guide for it OP
thumb.gif
Many thanks!

I started a guide for this way back in 2009 at ExtremeOverclockingForums.

This is very risky business. I didn't see another person actually be successful until 2014.

I may pop a tutorial to the boards here. It's quite a simple procedure really. The heating part only takes about 60 seconds and the solder is melted enough to remove the IHS plate. The rest of the tedious work is prepping and lapping the silicone free of solder.
 
#25 ·
Quote:
Originally Posted by ShrimpBrime View Post

Many thanks!

I started a guide for this way back in 2009 at ExtremeOverclockingForums.

This is very risky business. I didn't see another person actually be successful until 2014.

I may pop a tutorial to the boards here. It's quite a simple procedure really. The heating part only takes about 60 seconds and the solder is melted enough to remove the IHS plate. The rest of the tedious work is prepping and lapping the silicone free of solder.
Thanks looks like it'll be a good read. I'd be a little sketched out doing it with it being soldered but will eventually give it a go when I got extra cash laying around to order another one.
 
#26 ·
Would be glad to guide you all the way through the process. That guide is old, So the way I do it now differs from back then.

I have a little model vice that I clamp the PCB of the Cpu with and actually use a pair of ViceGrips to remove the IHS plate.

Here's a video. It's not of great quality, the camera bugged out. I've made a couple of other videos that probably are just as crappy lol.

FX-4300 - How I de-lid now
https://www.youtube.com/watch?v=B6ZNwVrhKkQ

Athlon 64 6400+ (soldered cpu) - How I de-lidded back then
https://www.youtube.com/watch?v=GjjXxHAhaG0

FX-4100 (failed de-lid)
https://www.youtube.com/watch?v=P4Hp0xQhJwg

The reason for the Vice and Vice grips is the glue cannot be cut completely as some of the little chips are glued to the IHS plate being very close to the edge there. So even with melted solder, almost have to force the plate off.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top