A hefty ATX PSU it seems:
https://smallformfactor.net/forum/threads/amd-project-quantum.99/#post-1338
Yeah, probably some passive 24V industrial powersupply in a fancy housing
-
Save 20% on ALL SFF Network merch, until Dec 31st! Use code SFF2024 at checkout. Click here!
Stalled kees KogelMier - a tiny powerhouse
- Thread starter QinX
- Start date
Yeah, probably some passive 24V industrial powersupply in a fancy housing
I have both I believe. Back when I started on H2O-Micro I contacted HDPLEX because of coilwhine coming from the HDPLEX when using a reference GTX970, after they investigated they sent me a new version to test and I think this is the V2.
V1 up top and V2 on the bottom I believe.
V1 up top and V2 on the bottom I believe.
Maybe you could try the dell 330w it is able to handle 420W. I test it a long time ago with an overclocked 2500k @ 4.5ghz and a OC GTX 780 with the HDPlex. Running Prime95 + 3DMark I had maximum values of 420W. Maybe it could go higher but at that time I had no stronger hardware.
Thanks for the suggestion, that might be a good idea to try with the HDPLEX and I might look into it. However for my build I'm only looking for 12V powersupplies. Although if people are willing to sacrifice the 2.5" mount the HDPLEX could fit, I'll have to check that out.
My #1 goal is to make a good uSFF entry level case that is still good looking. No expensive DC-DC board, but if people are willing to spend the money to get certain configurations working, why would I be against that. Although I'm starting to get the idea the R9 Fury might be a more stable choice then the R9 Nano
It seems absolutely crazy that the R9 Nano would peak so high over its power draw limits. Especially as this occurs mainly with low settings, you could try to contact AMD about it, maybe you've actually found some sort of bug in their firmware. Whether it is fixable or not is a different story.
I guess most people (i.e. other cases) won't be affected by this issue, though. SFX PSUs can easily supply the 470+ Watts peak you measured. It is very good to know if you want to go external PSU, though, maybe @confusis would like to make a short news article from these results?
I guess most people (i.e. other cases) won't be affected by this issue, though. SFX PSUs can easily supply the 470+ Watts peak you measured. It is very good to know if you want to go external PSU, though, maybe @confusis would like to make a short news article from these results?
I wouldn't be surprised this was normal behavior with other cards too. When I see my CPU flip from 100% usage at 65°C and a second later at 0% usage and 33°C (my G3258 shows this), I'm amazed how fast and drastic the dynamic power usage has become. PSUs are also able to sustain high peak usage without issue while some are even able to maintain much higher continuous usage like the ST45SF-G which went well over 500W sustained when cooled normally.
FurMark also has something to do with this as it is known for causing an unrealistic load, much like Burn-In tests with 100% AVX2 instructions. The 290X which is known for being power-hungry, typically loads for about 240-250W (100% not fps-capped) with peaks of about 270-275W, while FurMark showed 315W when stable, but peaks of around 400W during the start. (yes I love using colors).
Source: https://www.techpowerup.com/reviews/AMD/R9_290X/25.html
FurMark also has something to do with this as it is known for causing an unrealistic load, much like Burn-In tests with 100% AVX2 instructions. The 290X which is known for being power-hungry, typically loads for about 240-250W (100% not fps-capped) with peaks of about 270-275W, while FurMark showed 315W when stable, but peaks of around 400W during the start. (yes I love using colors).
Source: https://www.techpowerup.com/reviews/AMD/R9_290X/25.html
I'm more or less on the same boat as iFreilicht, mostly because I feel they way the R9 Nano is handling power management is just absurd.
That your CPU switches so fast is normal a GPU does that do. But what a CPU doesn't do is have a TDP of say 88W and pull peaks of 200W. This is what the problem is.
Yes if you load the CPU with AVX2 instructions you can have it run at 110W, but it will be stable at the wattage.
The GPU on the other hand is bending the rules quite a bit. They say the TDP of the GPU is xW but they will allow the GPU to take 2*xW or even 3*xW. As long as the average stays xW.
This may have been the case for a long time, but I feel this might be the first GPU to take it to this type of extreme, granted the card is targeted at an niche and I'm trying to use the card inside a niche of that niche.
I've not noticed this on the GTX970 which should also be a "noisy" card in terms of power consumption.
http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-12.html
Here is what happens in the span of 1 millisecond.
http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-11.html
Right now I'm running Heaven 1080p 0xAA and with the RSP-1000.
The RSP-1000 has a typical efficiency of 88%, but let us assume because of the low load that it is 82% efficient.
R9 Nano specification says typical board power of 175W.
Peak Power: AC-439W/DC-360W
Average Power: AC-220W/DC-180W ( RSP-1000 consumed 0.22kWh after running for Heaven 0xAA for 1 hour)
Monitoring the meter I see it ranging between between 215W and 260W with occasional dips/spikes to 160W and 330W
One of the key features of the card to achieve it's size, performance and noise levels is the power management. You'd imagine that would mean super tight control of the power consumption, but that only seems to apply to the typical board power not the maximum board power.
I am really tempted to order the DELL XM3C3 330W adapter just to see if the extra headroom, according to dondan, is just enough for it to run without problems.
Here is a really quick render I've made of the new design
That your CPU switches so fast is normal a GPU does that do. But what a CPU doesn't do is have a TDP of say 88W and pull peaks of 200W. This is what the problem is.
Yes if you load the CPU with AVX2 instructions you can have it run at 110W, but it will be stable at the wattage.
The GPU on the other hand is bending the rules quite a bit. They say the TDP of the GPU is xW but they will allow the GPU to take 2*xW or even 3*xW. As long as the average stays xW.
This may have been the case for a long time, but I feel this might be the first GPU to take it to this type of extreme, granted the card is targeted at an niche and I'm trying to use the card inside a niche of that niche.
I've not noticed this on the GTX970 which should also be a "noisy" card in terms of power consumption.
http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-12.html
Here is what happens in the span of 1 millisecond.
http://www.tomshardware.com/reviews/nvidia-geforce-gtx-980-970-maxwell,3941-11.html
Right now I'm running Heaven 1080p 0xAA and with the RSP-1000.
The RSP-1000 has a typical efficiency of 88%, but let us assume because of the low load that it is 82% efficient.
R9 Nano specification says typical board power of 175W.
Peak Power: AC-439W/DC-360W
Average Power: AC-220W/DC-180W ( RSP-1000 consumed 0.22kWh after running for Heaven 0xAA for 1 hour)
Monitoring the meter I see it ranging between between 215W and 260W with occasional dips/spikes to 160W and 330W
One of the key features of the card to achieve it's size, performance and noise levels is the power management. You'd imagine that would mean super tight control of the power consumption, but that only seems to apply to the typical board power not the maximum board power.
I am really tempted to order the DELL XM3C3 330W adapter just to see if the extra headroom, according to dondan, is just enough for it to run without problems.
Here is a really quick render I've made of the new design
Last edited:
I agree with everything QinX said, it is to be expected that a Processor peaks over its TDP, but not up to triple the Wattage.
After seeing the numbers of the GTX970, I'm quite impressed that my 300W PSU can actually handle Furmark and Prime95 on my PC at the same time.
The new design looks rather odd to me in terms of formfactor, but I just like the clean-ness of the front and top. Not sure if I said this previously, but I believe that punched round holes could be a bit cheaper to manufacture than these more complex laser-cut ones. If you're going to use a laser-cutter anyway, though, your design is the cheaper solution.
With this design, how much space is there below the case for the GPU to draw air from?
After seeing the numbers of the GTX970, I'm quite impressed that my 300W PSU can actually handle Furmark and Prime95 on my PC at the same time.
The new design looks rather odd to me in terms of formfactor, but I just like the clean-ness of the front and top. Not sure if I said this previously, but I believe that punched round holes could be a bit cheaper to manufacture than these more complex laser-cut ones. If you're going to use a laser-cutter anyway, though, your design is the cheaper solution.
With this design, how much space is there below the case for the GPU to draw air from?
The render doesn't do the new design really justice, I've just slapped some materials on there and threw into a scene. But what do you mean by odd in terms of formfactor?
A punched hole would be cheaper, my manufacturer has a single machine that does lasercutting and punching. I wanted to try something different and after the first prototype I will talk with my manufacturer about a dedicated punch tool.
Right now I've got 7.5mm high feet underneath the design so that would be the spacing below. I'll be testing out different heights once I have the prototype.
Thanks!
What I meant was that the proportions are unusual for a PC case. It is deep like a lot of HTPC cases and relatively slim, but it is so very narrow, unlike anything I've seen yet.
Oh I love those things, they are absolutely amazing machines.
Right now I'd say that is too little, but you never know, with a reference card it could actually work out quite well.
Hey I really like this. I love the look of your vent design. Personally I would love to see venting on the top as well, but I can understand why it is omitted.
Great job making it your own.
Great job making it your own.
Similar threads
