FX 9590??

[EarthDog]

No, it won't make it much if any cooler BUT it will help to stabilize temps/control temp spikes under load

I don't follow that logic. From my understanding of the physics of it all (which may not be correct) that makes no sense. There is nothing you can do about quick spikes. The heat doesn't transfer fast enough from the die to the IHS, through the TIM, and through the block to the water. Faster flowing water won't do anything because it is not what would prevent the spikes from happening. We already know from Skinee's/Martin's testing that 1-1.5GPM is the best and there are severely diminishing returns after that point. Only so much heat can transfer so quickly through all those mediums. Am I thinking about it wrong? Thanks for further explanation.

I am working with over 700W to dissipate in a loop of mine... one of which is a 500W part (R9 295x2). That said, I roll with 5x120mm total rads (2x120 is a double thickness) to dissipate that load. After the loop gets saturated, the GPU(s) max out at ~60C while the CPU (5820K @ 4.2-4.8GHz) hits 70C at the high end clocks (this is looping Heaven and running AIDA on the CPU). He doesn't have that much more to dissipate really... 300W CPU, 500W GPUs, and 50W chipset. Let's round up and say 900W worst case, that Phoyba should handle it, no? I do agree with rather have more than less, but...seems there is enough?

 

[DefiantReaper]

For the people telling me to go intel.. if youd re read my second post i had an i7 4790k build... and i plan on doing a i7 6700k build...
again this is my AMD build for ****s an giggles.. i wanna hit at least 5.1GHz is why i NEED to put the NB on a separate loop... i dont want the heat from the cpu dumping into it..
i probably will sell my 2 7970 GHz edition cards for a fury X... im just assuming the life of my 7970's will last MUCH longer with DX12.. my 3GB will then be 6GB because of SSR instead of CFX/SLI..
As for the motherboard i already have the top of the line is why im even considering a FX 9590.. i have the ASUS ROG Crosshair V Formula Z which always seems to get the best clocks...
+1 for Shrimpy.. you're absolutely correct..
just debating on a HEAVILY OC 8370.. a 9370... or 9590... and a 9590 is only $220 these days..

- - - Updated - - -

As for a good PSU i also have a Corsair AX1200i which i believe to be plenty well and stable..

 

[Dr. McCoy]

I don't follow that logic. From my understanding of the physics of it all (which may not be correct) that makes no sense. There is nothing you can do about quick spikes. The heat doesn't transfer fast enough from the die to the IHS, through the TIM, and through the block to the water. Faster flowing water won't do anything because it is not what would prevent the spikes from happening. We already know from Skinee's/Martin's testing that 1-1.5GPM is the best and there are severely diminishing returns after that point. Only so much heat can transfer so quickly through all those mediums. Am I thinking about it wrong? Thanks for further explanation.

I am working with over 700W to dissipate in a loop of mine... one of which is a 500W part (R9 295x2). That said, I roll with 5x120mm total rads (2x120 is a double thickness) to dissipate that load. After the loop gets saturated, the GPU(s) max out at ~60C while the CPU (5820K @ 4.2-4.8GHz) hits 70C at the high end clocks (this is looping Heaven and running AIDA on the CPU). He doesn't have that much more to dissipate really... 300W CPU, 500W GPUs, and 50W chipset. Let's round up and say 900W worst case, that Phoyba should handle it, no? I do agree with rather have more than less, but...seems there is enough?

My fault , I meant the amount of spike related to how high it gets when it does spike is better controlled as in these spikes aren't as severe when they top out. I agree you can't "Stop" these spikes so you have to deal with them as they occur.
Yes, there's only so much you can do about it since you are right about how quickly heat is passed through all the materials involved, I call that "Thermal Latency".

Obviously copper has a lower latency then aluminum does in how fast heat can pass through it. As for how fast the water flows, you can have it passing through so fast it actually doesn't get all the heat removed as it passes through the rad meaning it starts building heat, I've seen this happen with vehicles with the thermostat removed, it can only flow so fast through the rad for removing enough heat from the water. It also takes time for the water to absorb heat too, the longer it can stay in the block the more heat it picks up but it still has to be moving as in heated water going out, cooler water coming into the block for it to work or you're basically just running a miniature water heater. There's also a limit to how much a given amount or volume of water can take on related to heat, thats why it has to be moving.

The volume of water it gets is the big difference here, I've seen a difference between using 3/8 hose and 1/2 hose, redid my setup sometime ago and the severity of what temps were reached under load was improved after I did that. Using a larger size of tubing/hose with the correct fittings will help much more than simply having it flowing faster and as said you could have it flowing too quickly.
The ability to vary the speed your pump runs at is useful for tuning the system for maximum efficiency, in my case full-on works best but that's not the deal with all setups.

Skinee's/Martin's testing shows this plainly as to there's a limit to it all, the trick is to find it with the setup you have with all the factors involved. Maybe I don't have it all right but that's how it's worked for me and you guys have seen what I can do with watercooling so I must at least have it right to an extent... I guess.

 

[caddi daddi]

my phobya g-changer can cool my 8350 for about 48 hours at 5.5, with 8 r4 fans at 12 volts before temps get into the 50's, in a 75f room after that add about 2c for every day after.
my mora with 8 silverstones or my super nova with 16 r4's will cool it for days on end without issue, using either a raystorm or ek block, fans on the vrm's and a very low flow fan exhausting behind the socket.
this is all cpu only.

 

[EarthDog]

My fault , I meant the amount of spike related to how high it gets when it does spike is better controlled as in these spikes aren't as severe when they top out. I agree you can't "Stop" these spikes so you have to deal with them as they occur.
Yes, there's only so much you can do about it since you are right about how quickly heat is passed through all the materials involved, I call that "Thermal Latency".

Obviously copper has a lower latency then aluminum does in how fast heat can pass through it. As for how fast the water flows, you can have it passing through so fast it actually doesn't get all the heat removed as it passes through the rad meaning it starts building heat, I've seen this happen with vehicles with the thermostat removed, it can only flow so fast through the rad for removing enough heat from the water. It also takes time for the water to absorb heat too, the longer it can stay in the block the more heat it picks up but it still has to be moving as in heated water going out, cooler water coming into the block for it to work or you're basically just running a miniature water heater. There's also a limit to how much a given amount or volume of water can take on related to heat, thats why it has to be moving.

The volume of water it gets is the big difference here, I've seen a difference between using 3/8 hose and 1/2 hose, redid my setup sometime ago and the severity of what temps were reached under load was improved after I did that. Using a larger size of tubing/hose with the correct fittings will help much more than simply having it flowing faster and as said you could have it flowing too quickly.
The ability to vary the speed your pump runs at is useful for tuning the system for maximum efficiency, in my case full-on works best but that's not the deal with all setups.

Skinee's/Martin's testing shows this plainly as to there's a limit to it all, the trick is to find it with the setup you have with all the factors involved. Maybe I don't have it all right but that's how it's worked for me and you guys have seen what I can do with watercooling so I must at least have it right to an extent... I guess.

thanks for the details and clarification.

As far as the diameter of tubing, I'm not sure if you tested that empirically or if it was simply something you noticed when switching tube out. If it wasn't tested empirically, as you know there are several variables that come into play that could have given you that result (mount, tim application are two big ones). Skinee/Martin tested this empirically. The difference they found was negligible with the same flow rates with what I would guess to be repeatable testing methods.

A higher volume of water only means it will take longer to reach saturation/equilibium. As you well know, typical loops saturate in what, an hour or so assuming room temps remain the same. Someone above mentioned temps rise for days... that spits in the face of proven science so there are other factors involved I would imagine. If you put a constant heat load out, it shouldn't rise for days, all other external factors remaining the same.

 

[caddi daddi]

maybe bill nye, but after 2 days my temps rise about 2c per day, it's just more heat in the system than the rad and fans can move where it's installed, it's up against the wall behind my podium, the bigger rads have more surface area and can deal with the issues, heat input and the area it's installed in better, if I moved it out to the center of the room I'm sure the better airflow would cause the g-changer to keep up with it, it's just choked for fresh air.
as far as tube size, until you get to longer runs 3/8's is all that's needed.

 

[caddi daddi]

this is one 8 core fx cooling systems I tried, the fan you see blowing into the case is for cooling the vrm's, it works and I still do this.
the "thing" on the back of the case is a 30mm thick rad with three 12x38mmx3000 rpm fans on it.
even with these meat chopper fans it was not enough rad to go above 4.9 or so and was so loud that hartsfield airport in atlanta complained that I made to much noise.......

 

[ShrimpBrime]

For the people telling me to go intel.. if youd re read my second post i had an i7 4790k build... and i plan on doing a i7 6700k build...
again this is my AMD build for ****s an giggles.. i wanna hit at least 5.1GHz is why i NEED to put the NB on a separate loop... i dont want the heat from the cpu dumping into it..
i probably will sell my 2 7970 GHz edition cards for a fury X... im just assuming the life of my 7970's will last MUCH longer with DX12.. my 3GB will then be 6GB because of SSR instead of CFX/SLI..
As for the motherboard i already have the top of the line is why im even considering a FX 9590.. i have the ASUS ROG Crosshair V Formula Z which always seems to get the best clocks...
+1 for Shrimpy.. you're absolutely correct..
just debating on a HEAVILY OC 8370.. a 9370... or 9590... and a 9590 is only $220 these days..

- - - Updated - - -

As for a good PSU i also have a Corsair AX1200i which i believe to be plenty well and stable..

Since your really serious about the cooling, Just get the FX-9590. For a normal Joe with All in one cooling, I'd say get the 8370 and pray you can hit 4.7ghz stable..... LOL....

I already Know for a fact that 5ghz is going to require all of 1.5250v. All of it. Nothing less unless you get a better than average chip, and with degradation to keep in mind.... you'll likely need more than 1.5250v.

Now if you really want some decent overclocking pointers, I'd gladly help out. For example, raising Cpu VDDA 2.50000v to say 2.80000v is going to start pouring on heat when you combine it with CPU/NB and NB 1.8v. Hardly I see mention of the NB 1.8v, but believe me tread lightly where voltage is needed.

As far as the heat spike mentions occur, Your going to see this when that Cpu socket temp starts getting over 50c. At this point and beyond, this chip just gets warmer.

Pick and choose your stress testing battles. Most guys are really about Prime95. OK if it passes 2 hours, just keep in mind that doesn't mean it's gaming stable. It's calculation stable, meaning you'd probably be just fine Folding at Home after a Prime95 test. I always run calc testing, then some video rendering Like Unigene bench, and lastly real world gaming. If you can pass all that, it's going to be stable.

Truth about the heat. I'm running 120.2 X2 rads. It's not enough for the stock chip. Granted I have what I consider a crappy loop, old pump and other hardware.... but none the less what one would think is plenty of rad for just a cpu.... well it's not. Double that or more. Need a big res.

Ideas of my own - chilling 101

Was thinking on SS, but the power consumption is just no good. It would yield big time clocks on the other hand.

Then thought maybe a metal aluminum res with a bunch of peliters. Now here is less power consumption but possibly able to start fires if this isn't set up properly.

Used the blower.... It does squat.

Gotta chiller, but again like the SS not great for 2/4/7 use, at least the pelitier idea I can turn on and off.

But why the chilling you guys might ask?

Well because when I ran tap water to the drain on this processor, 5.1ghz stable is doable and the water temps aren't much warmer than 48f. So really, not much chilling is needed. Just enough however.... well ya over 5ghz can be had! It was some 5600mhz 1.6850v and Core temp hit a good 66c. Cpu socket temp was under 40c. When you see temps that far off, the heat transfer from core to the water block isn't really that great. This is why I de-lid. Get that waterblock closer to the heat source.

 

[Dr. McCoy]

As far as the diameter of tubing, I'm not sure if you tested that empirically or if it was simply something you noticed when switching tube out. If it wasn't tested empirically, as you know there are several variables that come into play that could have given you that result (mount, tim application are two big ones). Skinee/Martin tested this empirically. The difference they found was negligible with the same flow rates with what I would guess to be repeatable testing methods.

It was something I noted after having problems with some of my multi-cored chips not wanting to complete runs while benching. After I had switched things to 1/2 inch vs 3/8 hose it indeed made a difference. Yes these were mainly longer runs I was having trouble with.

A higher volume of water only means it will take longer to reach saturation/equilibium. As you well know, typical loops saturate in what, an hour or so assuming room temps remain the same. Someone above mentioned temps rise for days... that spits in the face of proven science so there are other factors involved I would imagine. If you put a constant heat load out, it shouldn't rise for days, all other external factors remaining the same.

I sometimes do runs without my rad fans running, just the pump moving water and I've seen this thermal saturation effect happen many times over, esp when I forget to re-enable the fans afterwards.
I normally do it for keeping air from moving through the rad when chilling it which would cause the rad to add heat, not remove it.
In this case it's better for fans to not be going for keeping temps low for a longer period of time vs with the fans going - Instead I just keep adding ice to the res as needed until I'm done with my run(s). Makes the ice last a little longer and it can hold temps a little better as long as ice is in the res itself.

My setup is made so I can add as much ice as I'd need, no worries about the res getting full after an extended period of chilling the water.

As for the rest this proves there is no set formula for doing a setup, all things considered there's just too many variables at work here. Hardware used, location, ect all add up to what's needed - Instead we can only suggest what's commonly known to have an effect towards the end we desire but nothing as an absolute to what you'd actually get.

 

[ShrimpBrime]

Ya know Doc, I've never used anything smaller than 1/2 OD hoses. Going bigger, I know makes very little difference. I figure the passageways in the waterblock are what matter most. Good flow through the water block is a must.

Either way, I hope DefiantReaper (Original Poster) comes back to tell us what he purchased and has some eye candies and such.

 

[bob4933]

I am considering getting an FX 9590 with the hopes of being able to hit at least 5.1GHz stable on all cores..
The reasons im wanting this processor is:
1: to support the under-dogs (AMD)
2: I'm guaranteed a 4.7GHz
3: I want the best i can get with AMD
4: i plan to HEAVILY WC my system already
5: I plan on doing a AMD themed PC (AMD RAM,SSD,GPU's,CPU, all fully WCed) Red
6: I already own the best board for this processor (ASUS Crosshair V formula Z)
I'm just wanting some inputs from people who own this processor.. I'm at a bottleneck with my 8140 and 2 7970's..
I also plan on water cooling my NB for extra stability on a separate 120mm loop.
THANKS!

1) foolish. Intel is superior in every way now...
2) not quite. Ive seen "unstable" (failing p95) 9590's stock.
3) not guaranteed. My 8320 hit 5.0 at 1.41v. That was p95 stable mind you.
4) then get an 83x0... save yourself some money for something you plan to do already.
5) thats a better reason than fanboi-ism
6) thats a great reason! You have the best am3+ motherboard... lead with that!





My 2 cents given.

 

[DefiantReaper]

bob... i find you quite ignorant with the intel... if youd reread my posts... i have stated that i had a i7 4790k build and sold it and now im waiting for the i7 6700k..
this is for an AMD themed pc.. for FUN (FUN!)
please keep fan-boyism out as i know intel is superior.
Now i plan on getting the Phanteks Enthoo Luxe and mounting a 420 on top and a 240 in the front for the 9590...
then i plan on adding a 120mm dedicated for the NB (high voltages)
then im going to add a 1080 for the graphics card setup i end up deciding to go...
(most likely Dual R9 Fury or Fury X)
i will definitely make a build log for this build..
and eventually for the Intel build around xmas..
then when im done with this system i will most likely sell it for a AM4 build early 2016 (tax time)
All rads will be painted red with ThermalTake Riing Red LED. (push on top 420) (Push/pull front 240) (push/pull rear 120 as exhaust)
I also plan on using the AMD Themed Swiftech Apogee XL on PPC's..
I greatly appreciate the feed back guys..
and bob i believe the FX 9590's are now more stable out of the box these days.. (ive seen it somewhere will post links when i find it)

 

[EarthDog]

The CPU was always stable. It was the boards an environment it was in that made those unstable. They do not make CPUs unstable at stock speeds.

 

[Johan45]

The CPU was always stable. It was the boards an environment it was in that made those unstable. They do not make CPUs unstable at stock speeds.

More or less, I've never had one but even S_B an experienced user was having some odd issues. He found I think in the end, the ONLY way that CPU was happy was with the sucker left on auto so it could tap the 1.5v+ even at lower speeds when it needed.
http://www.overclockers.com/forums/showthread.php/758169-Formula-z-and-FX-9590-Hangs-and-Freezing
This I think holds true from what I have seen in the forums as well. Most times once things in bios were being set manually people were having issues with hangs and freezing. It is probably one of the most demanding CPUs available right now and can be a real challenge once you stat turning the knobs.

 

[ShrimpBrime]

1) foolish. Intel is superior in every way now...
2) not quite. Ive seen "unstable" (failing p95) 9590's stock.
3) not guaranteed. My 8320 hit 5.0 at 1.41v. That was p95 stable mind you.
4) then get an 83x0... save yourself some money for something you plan to do already.
5) thats a better reason than fanboi-ism
6) thats a great reason! You have the best am3+ motherboard... lead with that!





My 2 cents given.

I'm curious about number two there. Got some links Mr. Bob4933??

More or less, I've never had one but even S_B an experienced user was having some odd issues. He found I think in the end, the ONLY way that CPU was happy was with the sucker left on auto so it could tap the 1.5v+ even at lower speeds when it needed.
http://www.overclockers.com/forums/showthread.php/758169-Formula-z-and-FX-9590-Hangs-and-Freezing
This I think holds true from what I have seen in the forums as well. Most times once things in bios were being set manually people were having issues with hangs and freezing. It is probably one of the most demanding CPUs available right now and can be a real challenge once you stat turning the knobs.

^^ Should be stickied IMO, lotta strong info there.....