Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Wanna TEC a Ryzen.
New Posts  All Forums:Forum Nav:

Wanna TEC a Ryzen. - Page 5

post #41 of 49
I applaud your push to TEC cool Ryzen, I believe there is a substantial clock gain to be had if one can just control the heat of the cores. Why I say this? well I can clock to 4.1gh with just a + 0.08vcore bump and it will run stable until the cpu hits 75C, then were done and it crashed. If I stop the test before the cores hit 75C and let them cool I can run the test again. If this was more of the mhz wall and not a heat wall I would end up with crashes when sub 75C core, but I don't.
I can raise the voltage but it starts taking lot's of voltage just to gain 25-50 mhz. and the temperature of the cores is spiking massively, like up to 105c running cpu-z's benchmarks. I've been overclocking AMD's cpu's for around 18 years now and It's all been higher temperature takes higher voltage to make stable but the higher voltage additionally causes higher temperatures. At my current maximum of 4175mhz I can nearly tie the 7700k on single thread and flat out destroy everything except Intel's 18core server chips on Multithread.
there is always naysayers and I listen to what they say but at the end of the day I usually end up proving them wrong via due diligence. My 8350 ran @ 5.2ghz @ 1.59vcore for years and I've never had a problem, no voltage degradation, no fried chips/motherboards, and after 3 years is still just works:thumb:
I'm currently working on the same thing and I've picked up a 62mm TEC and a 43 amp h-bridge to control the TEC off of motherboard's PWM fan controller. I'm not sure just how cold I'm going to take Ryzen but having a very custom water loop I'm not worried at all about cooling the 462watts of the hot side and holding it at ambient.
peltier. BTW while running Asus's 5 way optimization I saw my cpu use 240 watts @ 1.593vcore with the core speed of 4200mhz. the peltier/TEC I've bought claims it can remove 253 watts so I guess we'll find out:thumb:

https://www.amazon.com/gp/product/B01LYN0A94/ref=oh_aui_detailpage_o01_s01?ie=UTF8&psc=1
H-bridge
https://www.amazon.com/gp/product/B00WSN98DC/ref=oh_aui_detailpage_o01_s00?ie=UTF8&psc=1

http://s943.photobucket.com/user/seansplayin/media/4175_3200%20Multi%20thread_zpsp8ntlksp.jpg.html
http://s943.photobucket.com/user/seansplayin/media/4175_3200%20single%20thread_zpscbfx2mar.jpg.html
post #42 of 49
Quote:
Originally Posted by seansplayin View Post

I applaud your push to TEC cool Ryzen, I believe there is a substantial clock gain to be had if one can just control the heat of the cores. Why I say this? well I can clock to 4.1gh with just a + 0.08vcore bump and it will run stable until the cpu hits 75C, then were done and it crashed. If I stop the test before the cores hit 75C and let them cool I can run the test again. If this was more of the mhz wall and not a heat wall I would end up with crashes when sub 75C core, but I don't.
I can raise the voltage but it starts taking lot's of voltage just to gain 25-50 mhz. and the temperature of the cores is spiking massively, like up to 105c running cpu-z's benchmarks. I've been overclocking AMD's cpu's for around 18 years now and It's all been higher temperature takes higher voltage to make stable but the higher voltage additionally causes higher temperatures. At my current maximum of 4175mhz I can nearly tie the 7700k on single thread and flat out destroy everything except Intel's 18core server chips on Multithread.
there is always naysayers and I listen to what they say but at the end of the day I usually end up proving them wrong via due diligence. My 8350 ran @ 5.2ghz @ 1.59vcore for years and I've never had a problem, no voltage degradation, no fried chips/motherboards, and after 3 years is still just works:thumb:
I'm currently working on the same thing and I've picked up a 62mm TEC and a 43 amp h-bridge to control the TEC off of motherboard's PWM fan controller. I'm not sure just how cold I'm going to take Ryzen but having a very custom water loop I'm not worried at all about cooling the 462watts of the hot side and holding it at ambient.
peltier. BTW while running Asus's 5 way optimization I saw my cpu use 240 watts @ 1.593vcore with the core speed of 4200mhz. the peltier/TEC I've bought claims it can remove 253 watts so I guess we'll find out:thumb:

https://www.amazon.com/gp/product/B01LYN0A94/ref=oh_aui_detailpage_o01_s01?ie=UTF8&psc=1
H-bridge
https://www.amazon.com/gp/product/B00WSN98DC/ref=oh_aui_detailpage_o01_s00?ie=UTF8&psc=1

http://s943.photobucket.com/user/seansplayin/media/4175_3200%20Multi%20thread_zpsp8ntlksp.jpg.html
http://s943.photobucket.com/user/seansplayin/media/4175_3200%20single%20thread_zpscbfx2mar.jpg.html

You're gonna need a more powerful TEC, one of those won't be enough. It can move 253w to a zero degree delta, ie no cooling power, and because of all the added heat from the power consumption, you will be worse off than without the TEC because the hot side coolant will be well above ambient temp.

What water blocks will you use on the TEC?
post #43 of 49
Thread Starter 
I bought a combo from the Egg. A 1700, CrosshairVI, and 16 Gb PC3000 memory. This setup OC'ed to 4.1 Ghz stable in about 10 minutes. All I did was set 41x100, core volt to 1.4, soc to 1.15, and manually input the memory settings up to 2933. Have to learn a some more about all the new BIOS settings before I start doing some serious tweaking. I think I'm going to get the EK monoblock and set up a water chiller cooler when I upgrade my 2 290s for a Vega. Hopefully it will be out next month.
post #44 of 49
Be keen to know how 1700 goes with a bit of chilly coolant on it. Heard reports of 4.2 to 4.3 being about it but that is still a good gain.
post #45 of 49
Certainly not going to discourage the idea of building a chiller, because you can use it on anything else you want if the Ryzen turns out to be a disappointment.

But I'll throw out my $0.02. I've had a blast overclocking things like the Opteron 6276 (non-ES), Athlon II x4 640, as well as a few semprons where all you get is the base clock. On DICE, you'll get maybe +1GHz. That pales when compared to the FX-8350 where I got inches away from +2GHz (5.95GHz frown.gif ), but it was still enjoyable.

Now, the first processor that I've got where I quit overclocking it because it was a painful and fruitless experience was the 1800x. When you get an unstable overclock, the system continuously boot loops until you reset BIOS. That wouldn't be such a big deal if an unstable overclock wasn't 25MHz to 50MHz above 100% stable. When coming to this, I was used to the 8350, where you have a gradient from stable to unstable and you can land somewhere in the middle where doing too much will crash the system, but otherwise it boots. This is nonexistent with Ryzen. It either works fine or it's not stable enough for a validation and you will be spending the next several minutes waiting on a BIOS flashback.

So when trying DICE and in my chillbox, I'd guess that 50% of my time was spent sitting there waiting on BIOS to be reset. This entirely killed the experience for me. Maybe I was unlucky and you will have better results - but I have seen this now on multiple Ryzen CPUs as well as heard OCer's complain about similar things.


I could validate at 4.3GHz in the chillbox (subzero coolant).
I didn't stress test, but I'd guess it would be a world of pain if I did. This was +100MHz over watercooling results.
Edited by Skyl3r - 6/2/17 at 6:38am
post #46 of 49
Thread Starter 
Once I get a monoblock and a watercooled Vega I'll be in a better position to decide how and if to move forward with a TEC. IT does seem that there is a very finite ceiling with the R7 of 4.3Ghz. I'm not a bencher, I only use my computer a few evenings a week for gaming. I want a high stable OC that I don't have to fool with. I was hoping for a 5.2Ghz like my 8350 but even at 4.1 the 1700 blows the lid off the 8350.

That reminds me. I have to put the 8350 setup on Ebay. I will probably put it in the classifieds here first with the MB blocks installed.
EDIT: I didn't know you had to have 35 reps to sell here. Bummer. I'll check overclockers. I think you only need 100 posts...
Edited by Thick8 - 6/2/17 at 3:00pm
post #47 of 49
Quote:
Originally Posted by LiamG6 View Post

You're gonna need a more powerful TEC, one of those won't be enough. It can move 253w to a zero degree delta, ie no cooling power, and because of all the added heat from the power consumption, you will be worse off than without the TEC because the hot side coolant will be well above ambient temp.

What water blocks will you use on the TEC?
I looked into it a little bit more and your completely right, It's going to take a much more powerful TEC. I've found a couple rated a little higher @ like 378 watts like the ATE1-288-20AS, in looking at the chart here http://www.analogtechnologies.com/document/ATE1-288.pdf it looks like I can move 250 watts at a 25C temp drop at near maximum volts (36v). still kind of thinking I need a bigger one still, and this one is designed for 36volts which means I need a special power supply and cannot use an ATX.

anyhow you were totally right.

I'm an HVACR technician and I could easily build a super chilled phase change loop but I just don't like the idea of my computer being teathered to a condenser, the idea of everything inside the case appeals to me.
post #48 of 49
Quote:
Originally Posted by LiamG6 View Post

You're gonna need a more powerful TEC, one of those won't be enough. It can move 253w to a zero degree delta, ie no cooling power, and because of all the added heat from the power consumption, you will be worse off than without the TEC because the hot side coolant will be well above ambient temp.

What water blocks will you use on the TEC?

as for the TEC water blocks I really haven't had much luck finding copper ones and I don't think I can use Aluminum where the radiators and EK cpu water block are copper.
post #49 of 49
Quote:
Originally Posted by seansplayin View Post

I've found a couple rated a little higher @ like 378 watts like the ATE1-288-20AS, in looking at the chart here http://www.analogtechnologies.com/document/ATE1-288.pdf it looks like I can move 250 watts at a 25C temp drop at near maximum volts (36v).

That's 50mm^2. There are stronger TECs in 62mm^2. In 50mm, there's also this one from Thermonamic.
Quote:
Originally Posted by seansplayin View Post

I'm an HVACR technician and I could easily build a super chilled phase change loop but I just don't like the idea of my computer being teathered to a condenser, the idea of everything inside the case appeals to me.

Yeah, that's it exactly. It should all fit inside the case and run quietly and not involve a reciprocating compressor. I'm working on a power supply to drive these big voltage TECs in the case. It's slow going because I have a day job and 4 other projects going at the same time. But I intend to be able to deliver 850watts at up to 36Vdc. I might try to push a bit farther than 850 but that's what I'm shooting for now. That would be 23.6 amps at 36V max. My supply takes its input power from up to 3 8-pin PCIe power connectors from the main supply.

I'm hoping someone makes a direct die block for ThreadRipper myself. That package is large and would probably be more efficient at moving heat to a 62mm^2 TEC than the Intel stuff.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Peltiers / TEC
Overclock.net › Forums › Cooling › Specialized Cooling › Peltiers / TEC › Wanna TEC a Ryzen.