First manual overclock. FX-9590 Sabertooth 990x R2.0

[ShrimpBrime]

Stability testing nowadays takes ~20m on IBT AVX very high 10 runs for any non anal everyday overclocker.

Again nowadays getting to 4.7ghz is fairly easy to most 83** on air, passing that is the tricky part and for most cases will require an AIO or custom loop (I pull 4.7ghz ~55c, 4.8ghz ~60c on a Noctua NH-D15 on all 8 cores).

AFAIK and was told by the almighty's here, the 95** is simply a better binned overclocked 83**, so I assume that at 4.7ghz both CPU's will do roughly the same on any bench ?

Following that reasoning, can we run the 9590 on air at 4.7ghz with the same Noctua or do you need water ? This is the crux of the matter, correct ?

If this post had made better sense in light of what "Binning" actually means, it would produce some real results to the argument that a FX-8350 is essentially the same thing as a FX-9590.

Which it's not.

AMD tests these processors. For multiple things, one being stability at given x clocks and x voltage. Since FX-8350's didn't meet requirements to be "stable" at 4.7ghz.... they where binned much lower. Thus the need for testing. Essentially Our testing doesn't meet manufacturing requirements on what Silicon will do stable.

We can test for stability to liking. 20 minutes IBT is a Joke in my book, not meeting my personal requirements of a minimum of 8 hours stable crunching P95 and OCCT Linpack back to back and then a couple hours worth of gaming after wards. I even consider 2 hours P95 stable somewhat of a Joke. It's only running a calculation stressing a portion of RAM a lot of CPU resources and nothing more.

Back in the day, PiMod could take hours. That was true stability testing as single cores where ramped 100% during the test.

(I pull 4.7ghz ~55c, 4.8ghz ~60c on a Noctua NH-D15 on all 8 cores).

What temp? Core temps? Cpu socket temps? Is this only 20 minutes? That's quite a short test to boast stability.

Following that reasoning, can we run the 9590 on air at 4.7ghz with the same Noctua or do you need water ? This is the crux of the matter, correct ?

AMD states themselves to Use liquid cooling. Since AIO is soo mainstream even possibly over air cooling now, the question is null.

The issue is people expect to overclock the chip while it's already at that speed your Noctua may or may not handle under x and y circumstances. But then I hardly believe a Noctua can handle any FX processor at 1.5v and 5ghz any ways, the stretch is always for a higher mhz.

There's only ONE FX-8320 that I am aware of stable at 5.2ghz. Mr. Scott has it and is de-lidded as mine is. For the sake of truth, that Cpu is not stable on ambient 75f air temps on liquid cooling. I should know, it was tested with the same board he was also sent with. I had a CB issue with bios, he had overcome with a bios tweak of some sort. No business of mine what's done with after it leaves my hands, but I know the truth on most FX processors.

Phenom II hit or miss 4ghz stable and up and FX hit or miss 5ghz stable and up. Depends on leakage, cooling, voltage used.... but in most cases can replicate any ones overclock with similar cooling and clocks/voltage with my very own setup.

But because we are talking about stock processors and then overclocked...... My FX-9590 with the mods and adequate cooling boasts clocks at 5.2ghz daily. 20 minutes anything os no problem. Game for 8 hours straight no problem.

Now take look at Mr. Eagle there. Great guy!" Always in search for a better temp applying all kinds of modifications and getting results in the meantime.

However, if we take a stock FX-8350 and pit it up against a stock FX-9590, set cooling aside, the FX-8350 is a pony.

In the end, it's always a struggle with cooling... Your Max clocks are still 200mhz less than a stock FX-9590 turbo frequency.

You stability test.... I go off gaming.

Here is highest reported Liquid cooling speed at HWBot with FX-9590 http://hwbot.org/submission/2784297_shrimpbrime_cpu_frequency_fx_9590_5718.05_mhz

Here is highest reported clocks at HWBot with FX-8350 liquid cooling http://hwbot.org/submission/2790171_tt_chuchu_cpu_frequency_fx_8350_5709_mhz

The 8350 used less voltage but gained no more mhz than the 9590. What gives? The 8350 had tens of submissions while the 9590 had maybe 10 submissions on the same cooling. Comparison is null.

 

[Kenrou]

If this post had made better sense in light of what "Binning" actually means, it would produce some real results to the argument that a FX-8350 is essentially the same thing as a FX-9590.

Which it's not.

AMD tests these processors. For multiple things, one being stability at given x clocks and x voltage. Since FX-8350's didn't meet requirements to be "stable" at 4.7ghz.... they where binned much lower. Thus the need for testing. Essentially Our testing doesn't meet manufacturing requirements on what Silicon will do stable.

We can test for stability to liking. 20 minutes IBT is a Joke in my book, not meeting my personal requirements of a minimum of 8 hours stable crunching P95 and OCCT Linpack back to back and then a couple hours worth of gaming after wards. I even consider 2 hours P95 stable somewhat of a Joke. It's only running a calculation stressing a portion of RAM a lot of CPU resources and nothing more.

What temp? Core temps? Cpu socket temps? Is this only 20 minutes? That's quite a short test to boast stability.

The issue is people expect to overclock the chip while it's already at that speed your Noctua may or may not handle under x and y circumstances. But then I hardly believe a Noctua can handle any FX processor at 1.5v and 5ghz any ways, the stretch is always for a higher mhz.

The 8350 used less voltage but gained no more mhz than the 9590. What gives? The 8350 had tens of submissions while the 9590 had maybe 10 submissions on the same cooling. Comparison is null.

My temps are overnight Prime95 blend (at least 8 hours since me and wife go to bed rather early), and the IBT like i said is for non anal, quick test for stability. This is a catch 22, you can stress for 24h straight and be all good and you can say "hey it passed" just to see it fail at 24h01m BECAUSE you have been stressing him so much correct ? Before i started reading and posting in this forum i used to do Prime95 small fft 20m between voltage/speed bumps and overnight for final stress testing and always suited me just fine, never had problems with overclocking. Was told to use IBT simply because it was faster, and now i can just run the test like i mentioned and game/do video coding all day long without any issues.

Temperature should always be read at package, don't even bother with others unless i see a unreasonably high number.

Again between 83** and 95** i can only go through what i was told on other threads here, hence the question, but you showed yourself that they are matched but seemingly the 8350 got there using less voltage and i am assuming less heat, which is the whole point here. Suppose you could attribute that to chip maturity, but with everyone saying "Don't go with 95** unless you have MAJOR cooling" you will never see many submissions.

So my question stands : in terms of efficiency, the 8350 (assuming you can get it at 4.7ghz at a reasonable voltage AKA decent chip) is still better then the 95** correct ? mostly because the average person doesn't sell its house to buy a custom loop, at most they will do like me and get the "best" air cooler that fits in the case or a good AIO.

 

[ShrimpBrime]

My temps are overnight Prime95 blend (at least 8 hours since me and wife go to bed rather early), and the IBT like i said is for non anal, quick test for stability. This is a catch 22, you can stress for 24h straight and be all good and you can say "hey it passed" just to see it fail at 24h01m BECAUSE you have been stressing him so much correct ? Before i started reading and posting in this forum i used to do Prime95 small fft 20m between voltage/speed bumps and overnight for final stress testing and always suited me just fine, never had problems with overclocking. Was told to use IBT simply because it was faster, and now i can just run the test like i mentioned and game/do video coding all day long without any issues.

Temperature should always be read at package, don't even bother with others unless i see a unreasonably high number.

Again between 83** and 95** i can only go through what i was told on other threads here, hence the question, but you showed yourself that they are matched but seemingly the 8350 got there using less voltage and i am assuming less heat, which is the whole point here. Suppose you could attribute that to chip maturity, but with everyone saying "Don't go with 95** unless you have MAJOR cooling" you will never see many submissions.

So my question stands : in terms of efficiency, the 8350 (assuming you can get it at 4.7ghz at a reasonable voltage AKA decent chip) is still better then the 95** correct ? mostly because the average person doesn't sell its house to buy a custom loop, at most they will do like me and get the "best" air cooler that fits in the case or a good AIO.

My custom loop has served me since socket 939 days with things added and replaced along the way.

OK, at 4.7ghz loaded FX-9590 typically runs about 1.4250v. At 5ghz, 1.5250v -

Should a person want to try running these clocks with ANY FX processor, A custom loop would be advised.

Temperature should always be read at package, don't even bother with others unless i see a unreasonably high number.

No, you don't seem to follow my posts here much (not talking strictly to this sole thread...)

85c Core temp is thermtrip design AMD processor limit.

CPU temp or SOCKET temp is a motherboard reading and reflects the base where the processor is plugged in. Heat is accumulated in Cpu to motherboard PCB to VRM back to CPU.

All temps should be watched.

Socket to save face on the VMRs and motherboard and Core to prevent thermtrip.

Most users encounter stability problems from the socket temp (motherboard) and thus loose stability here but is not a direct reflection of the actual processor reading.

Here -----> Start Page three post 57 and on. http://www.overclockers.com/forums/showthread.php/728489-FX-its-tougher-than-you-think/page3

It's not that this chip or that chip is better.... It's who's cooling capabilities are better.

For instance, My FX-9590 is limited by THERMTRIP of 85c. I have not encountered a stability problem while I can control my CPU (socket temp) rather well being the processor is de-lidded. So at about 5.4ghz I have a shut down issue exceeding THERMTRIP (85c) while my socket temp rarely would ever break 55c.

FX is not voltage limited. It's cooling limited. No Cold Boot / Cold Bug. Asus Crosshair Formula-Z has a max voltage limit of about 2.7v (I'd have to double check this for backed truth and I will shortly after I'm done typing). EDIT: Cpu v-core max is 2.3v via bios, I just checked.

Since we are not voltage limited, ALL FX PROCESSOR can clock to where ever you like given the supplied cooling is available. Any person that benchmarks extreme cold know this. The only Voltage limitations available are the VRM and PSU.

So what's better? A FX-8350 or a FX-9590. Well the FX-9590 binned at 4.7ghz 5ghz Turbo at stock with high leakage of course. If you seek a higher clock speed, you should seek better cooling, I don't really care what processor you have.

On a side note and testing low voltage with 0 to sub zero cooling, I've done this testing with a simple TEC. I had used the idea to keep frozen while cranking up CPU speed. In the meantime it was to see how far a cpu could go even if shutting down cores was needed and I obliged to this fact of lower TDP throughput to achieve the frozen effect and low voltage. You can view this demonstration here at this link I'll provide you now. http://classicplatforms.com/main/forum/viewtopic.php?f=200&t=8319

I would suggest signing up at CP (classicplatforms.com) to view the pictures.

So any how.... Saying FX-8350 is better than FX-9590 is to the like of saying FX-4300 is better than FX-8350 because it runs cooler. It just doesn't make sense.

 

[ShrimpBrime]

Here is P95 for 5 minutes 1.536v at load 1.5250v idle. I'd test longer but already have, temps won't go much higher. P95 set to maximum heat less ram used.

This FX-9590 humps harder than a horney monkey on a wedding night.

 

[Kenrou]

No, you don't seem to follow my posts here much (not talking strictly to this sole thread...)

Sorry i don't follow any user/thread unless i am thoroughly interested in the subject

85c Core temp is thermtrip design AMD processor limit.
CPU temp or SOCKET temp is a motherboard reading and reflects the base where the processor is plugged in. Heat is accumulated in Cpu to motherboard PCB to VRM back to CPU.
All temps should be watched.
Socket to save face on the VMRs and motherboard and Core to prevent thermtrip.
Most users encounter stability problems from the socket temp (motherboard) and thus loose stability here but is not a direct reflection of the actual processor reading.
Here -----> Start Page three post 57 and on. http://www.overclockers.com/forums/showthread.php/728489-FX-its-tougher-than-you-think/page3

I posted there almost 2 weeks ago and i am still waiting for a definitive answer to my question. I know all of this up, my socket is usually ~5c/~8c higher then package so well within the "72c threshold". Like i said don't even bother with others unless i see a unreasonably high number.

So what's better? A FX-8350 or a FX-9590. Well the FX-9590 binned at 4.7ghz 5ghz Turbo at stock with high leakage of course. If you seek a higher clock speed, you should seek better cooling, I don't really care what processor you have.

Why would it be better ? im missing something im sure, you just told me any FX depends on cooling, so a 8350 with the same cooling as a 9590 would provide roughly the same results. I asked this at the start : is the 9590 just a superclocked 8350 ? i meant it as, is the chip architecture the same ? did they simply pick better samples that offered the best overclocking results and called it a 9590 or it it a different beast altogether ?

So any how.... Saying FX-8350 is better than FX-9590 is to the like of saying FX-4300 is better than FX-8350 because it runs cooler. It just doesn't make sense.

I think this was meant as better overall, as in, same results with less voltage/heat/cheaper, not necessarily better in benchs. I said that the average user will use a good air cooler or a decent AIO, not a full custom loop like you guys, so the 5ghz turbo is beyond reach and he will have to stick with the undervolted 4.7ghz. In that situation it is better (in my idea) to buy the 8350 which is about £50 cheaper and will do the same job no ?

 

[ShrimpBrime]

I suppose you are right. I don't feel like skinning a cat. I've owned all the FX processors except the "e" series chips.

You saved 50$ and stuck with air cooling. OK. So that leaves you with a slower processor either way. You are roughly 500mhz behind my stable clocks on a custom loop with a high leaker chip.

In reality, no one needs more than a quad in most cases. So deciding to get an 8 core is just to say you have an 8 core. You could be 300mhz higher with 4 cores and saved 80$.

What was the point again?

Its a race. That's what hwbot was designed for.....

 

[ShrimpBrime]

Oh and let me answer your question.

Over clocking aside.... the fx-9590 was designed at stock and binned as such to have higher clocks than a fx-8350.

If motherboards where locked... hmm say buying a dell like an average user would.... to have a better processor ... keep in mind not over locking it via lack the ability too... the fx-9590 is a faster processor and was designed as such.

You pay more for a v8 and get the low end tourqe. Sure you could cam it for more HP and tourqe but decided to get the economical v4 and save fuel.

Hopefully you follow?

You can mod the v4 to have as much HP and tourqe as the v8 but need to spend the time and money to do so..... might as well just bought the v8 and mod that instead.

Hopefully you are a car guy somewhat and follow the drift lines and stay on course here..... ????

 

[Kenrou]

You saved 50$ and stuck with air cooling. OK. So that leaves you with a slower processor either way. You are roughly 500mhz behind my stable clocks on a custom loop with a high leaker chip.
Its a race. That's what hwbot was designed for.....

I'm not racing, just tinkering, i bought this octa cause besides games i do a lot of video editing and newbie me i though it would give me a leg up when DX12 came and the prices they set for parts/full custom loops... If i was single i probably would get one tbf but i got other priorities so i will stick to my Noctua tyvm (the 980 Ti alone cost more then my wife's car and she will castrate me if i ever spend so much money on my rig again)

Oh and let me answer your question.

Over clocking aside.... the fx-9590 was designed at stock and binned as such to have higher clocks than a fx-8350.

If motherboards where locked... hmm say buying a dell like an average user would.... to have a better processor ... keep in mind not over locking it via lack the ability too... the fx-9590 is a faster processor and was designed as such.

You pay more for a v8 and get the low end tourqe. Sure you could cam it for more HP and tourqe but decided to get the economical v4 and save fuel.

Hopefully you follow?

You can mod the v4 to have as much HP and tourqe as the v8 but need to spend the time and money to do so..... might as well just bought the v8 and mod that instead.

Hopefully you are a car guy somewhat and follow the drift lines and stay on course here..... ????

Right, better binned, got ya (not a car man myself but i understand the analogy). I stated that this was what i was told in my 1st post and you seemed to disagree with it ?

 

[ShrimpBrime]

Its when we use the words average user in a OC forum that the disagreement comes the FX 8 series is just as fast.

We the maybe one percent of PC users actually modding and overclocking are the only people that could ever try and say a 8350 is just the same as the 9590 but only because we had to over clock it.

Other than technicalities they are essentially the same processor design, the difference is the 9590 will never post below 4.7ghz even when you clear cmos which is not the case with a fx-8350.

Clear cmos and your back at 4.2ghz turbo at Max.

That's what I mean about the 9590. Its never going to be slower than a 8 series in any way shape or form while setting cooling aside.

I recommend water cooling any FX processor for people that are serious about keeping a decent temp. Then again most gamers just want to settle for a little more vs maxing capabilities like you and I have with our cooling solutions.

 

[Johan45]

I just saw something interesting about the 9590 chips from the STILT. I always knew they were high leakers hence the power and heat they produce but what he says is almost ironic considering they actually charge more for them

[quote name="The Stilt" url="/t/1318995/official-fx-8320-fx-8350-vishera-owners-club/55930#post_24688379"][quote name="KyadCK" url="/t/1318995/official-fx-8320-fx-8350-vishera-owners-club/55900_50#post_24688186"]
A 9590 would never run at 4Ghz anyway, irrelevant. AMD's power saving states are not accurate enough to bother with any of them, they just jump from minimum clocks to max turbo with any real load.

Irrelevant, how? Those are the figures the chip is rated to, which are just as valid as the nominal TDP rating.

The whole point being that no matter how much you underclock or undervolt the FX-9K parts, they will always consume significantly more power at the same settings as an equivalent FX-8K series part at the same clocks. All FX-9K parts have abnormally high SIDD, which is the reason AMD was able to release them clocked so high. Eventhou the equally binned normal silicon, with normal leakage characteristics could do the same clocks at lower power consumption, they couldn´t do it with voltage levels AMD would consider safe.

Unlike the end-user, AMD must take account the load-line specifications built in the infrastructure specifications and so on. If a chip with normal leakage characteristics requires 1.45V in order to pass AMD qualifications at 4700MHz, they would have to rate it to ~1.625V default voltage just to comply with their own specifications (1.3mOhm RLL).

The FX-9K series is an abomination and if it wouldn´t exist, the dies now sold as FX-9K series would have been scrapped before even leaving the Dresden fab.

It's almost funny him and Kyad are kind of getting into it over this at OCF http://www.overclock.net/t/1318995/official-fx-8320-fx-8350-vishera-owners-club/55940

 

[Blaylock]

LMAO

 

[ShrimpBrime]

We already know they consume more power.... says that right on the box.

The deal is as boxed as all trying to be said that the 9 series chips are faster from the box period.

When you oc fx to 4.7ghz, its going to be roughly the same tdp.

At that point you can call it the same I suppose minus the fact when the of goes bad and you clear cmos you have a stock stable chip again.

But to say that because you tested your Fx 8 series at 4.7ghz and it passed your testing does not make it a binned 4.7ghz equivelant to fx 9590. If this is true, mines be binned at 5.2ghz where its kept daily upon my own testing which may or may not be good? Enough at others discretion.

Its neither here nor there.

End result.....

You can't cool it... you can't clock it. Even Stilt hasn't an argument for that statement, cause I've already asked or stated it. There was nothing more to be said.

 

[Kenrou]

It's ok, i am genuinely the last person here wanting to start an fight, i just want to understand where it stands. I thought maybe the 9590 was a another class of CPU and have differences like the Bulldozer from the Vishera, but everyone kept telling me it was just a better binned 8350. Stilt says even though its higher clocked/binned it produces more heat comparatively, which was the statement that started the argument.

Again this CPU seems to be out of reach for most people because it requires major water cooling right ?

 

[ShrimpBrime]

Right. It requires water cooling. Has high leakage. Is not intended to run below 4.7ghz and is AMDs fastest boxed processor.

I do not recommend AMD to average consumers actually. To over clockers I do.

The 9590 runs hotter than any other chip on the market but also boxed faster then the rest of AMDs line up.

Marketing has it stated a 5ghz CPU. I see 5ghz all the time when it turbos. I don't have to change not one setting in bios to accomplish that speed.

I enjoyed the back and forth and do not consider it an argument. I'm not upset in any way either.

If you want high percentage oc, you buy fx-4300 or fx-8300. My 8300 did over 100% overclock and it was great fun. But at 3.3ghz boxed in comparison to fx-9590 or even fx-8350.... it was considerably slower because with fx processors, every MHz counts.

And about gaming at 7ghz, the frame rate was about 2% higher than running around 5ghz. But then again this is a guesstimate because I didn't do some full on testing, I just simply gamed for 20 minutes just to say I've done it. No different than a bloke bragging a cpuz round here

 

[Aval0412]

i have the same chips fx9590 with asus sabertooth fx990 rev 2.0 running @5100.34 voltage set to 1.46 and never experienced temperature rising high and hot.prime 95 blend test my temps only reached 52 the most.still perflexed about this heat thing on fx 9590 by the way my cooler corsair h110i and psu 1300watts evga super nova.im a new member you can check my profile

 

[Kenrou]

FX-9590 @5.1ghz 1.46v...

Yeah... No. Just no. Just... No... Maybe... No. No dude, not happening

 

[Mr.Scott]

Sometimes you find a lump of gold even in the biggest pile of crap.

 

[Aval0412]

check my cpu-z validation and im going to upload screeshots of prime95 blend testing

 

[Kenrou]

Sometimes you find a lump of gold even in the biggest pile of crap.

That's not gold, its diamond

 

[Mr.Scott]

That's not gold, its diamond

True.....but still possible. Not the norm at all, but possible.