RTX A2000-X3 | a cool&quiet tripple fan mod for the A2000

Snerual

Cable-Tie Ninja
Jul 3, 2020
196
224
What is power draw like with such a shunt mod? I would be afraid to fry my motherboard... I've also seen people solder a cable to the shunt and then attach a PCIe 6 pin that way. I assume that then evenly spreads the load between the motherboard slot and the extra cable? (i.e. if you are drawing 80 watts it will be 40 from the slot and 40 from the extra cable)
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,166
3,504
www.revoccases.com
No, for maximum powaaaa you need to do the shunt mod on the front and the back of the PCB. The additional +12V cable shall just prevent that you fry your PCIe slot / motherboard
 

Snerual

Cable-Tie Ninja
Jul 3, 2020
196
224
I spent quite some time searching for information online about shunt modding but... it's hard to get some real conclusive answers and fully understand the consequences. Some people add a 5 mOhm resistor only to the top shunt, others do it on both. Other people claim that is "too much" and in stead replace the existing shunt with a 4 mOhm one (also either on only one shunt or on both...) Even with adequate cooling and a beefy motherboard... doesn't the 3 phase VRM of the GPU become an issue? I can't imagine that's designed for 100W+ of power...

I also really don't get the 12V cable mod... sure, in theory now the GPU draws half the power through that cable, but what about ground? Your only ground connection is still via the PCIe slot... And if ground is irrelevant in that case, why do 6 and 8 pin PCIe power connectors contain so many ground pins?

BTW: don't be sad about your 3DMark GPU score... I can't get mine above 6000 no matter what I try...
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,166
3,504
www.revoccases.com
most miners add one 5mOhm resistor in parallel / on top of the front one, AFAIK this is considered to be the "safest" mod that won't fry your GPU and motherboard (this is the mod I did and my card wouldn't draw more than 76W peak during benchmark)

with the other mod which adds 5mOhm resistors on both sides, plus the +12V cable, the A2000 could draw up to 140W and I agree this certainly isn't working well for the VRMs long time... IMHO this is more for overclockers who want to beat @SFFMunkee 's Timespy score, not so much for everyday usage XD

concerning the +12V cable: you don't need an additional GND cable because the PCIe slot has far more GND pins than +12V

I can't get mine above 6000 no matter what I try...

how's your voltage / clock curve looking?
 

SFFMunkee

Buy first, justify later?
Gold Supporter
Jul 7, 2021
944
1,012
For what it's worth, I've shunt-modded my A2000 (front and back additional 0R005 AKA 5milliOhm) and I'm still using slot-power (albeit from a separately powered riser). I do NOT recommend running this at full tilt (or from slot-power) 24/7 as you'll risk cooking the power circuitry or frying the PCIe slot on your motherboard.

If you haven't shunt-modded, you'll want to keep your frequency/voltage curve fairly conservative (I had max 750mV at 1590MHz) as the power limit is pretty restrictive.

You can always try doing a single 5mOhm shuntmod first, then add the second if you're not happy with the results. Remember that with the mod, and thus higher power limit, it opens up the higher voltages so you can get MUCH higher clock speeds.

As a reference, here's a quick 'n' dirty comparison based purely on my online TimeSpy results:
1650MHz - Average 1505MHz STOCK
1792MHz - Average 1776MHz 1xSHUNTMOD (front 0R005 in parallel)
2145MHz - Average 1961MHz 2xSHUNTMOD (front AND rear 0R005 in parallel)

From memory, I had 750mV on stock, about 975mV 1xSHUNT and 1025mV or something for 2xSHUNT.
 

SFFMunkee

Buy first, justify later?
Gold Supporter
Jul 7, 2021
944
1,012
You can use this handy utility to calculate the difference in power limit. https://github.com/bmgjet/ShutMod-Calculator (please note: this is not mine, and I take no responsibility and make no guarantees to the code/executable provided).

But theoretically, if you halve the resistance on the shunt then you effectively double the limit. Likewise double the value shown in HWInfo or GPU-Z to get the 'true' (still estimated) power usage.

5mOhm (stock) + 5mOhm (mod) = 2.5 mOhm --> 140W (power readings x2)
5mOhm (stock) + 4mOhm (mod) = 2.22mOhm --> 157.5W (power readings x2.25)

My assumption is that the circuitry uses both the front and rear shunt resistor, and averages them (or something similar), hence modding only a single resistor is less effective than doing both.

If you wanted to be more cautious, you could use a slightly higher value resistor (6-10mOhm). I stuck with the 5mOhm as it made it a nice round, easy to calculate number and also was tested and proven. :)

::EDIT:: I hadn't realised quite how few 12V pins there were compared to GND.

A PCIe x16 slot has only FIVE 12V pins, but SIXTY-EIGHT GND pins! (x8 slots have 37, x4 slots have 21, x1 have 9)

75W / 5 = 15W per pin = 1.25A per pin
140W / 5 = 28W per pin = 2.34A per pin

Interesting.
 
Last edited:

Snerual

Cable-Tie Ninja
Jul 3, 2020
196
224
Very interesting information, thanks!

My curve currently looks like this



But during the benchmark it never exceeds 1250 MHz or so... keep in mind I am running with a 97% power limit because my PSU is still causing occasional shutdowns at 100% (a beefier one is on the way from China...)

However Afterburner also never shows the current mV, is this normal?



I assume my B550 Aorus is strong enough to handle the "light" shuntmod without extra 12V cable. Speaking of the 12V cable, I also feel like it's a bit weird to just directly pipe an unfiltered 12V line from your PSU in parallel to 12V from the motherboard that traveled through a bunch of other components and probably has a very different looking ripple profile... I assume GPUs with proper PCIe power connectors don't do this the same way but also apply some filtering before combining all the 12V lines... but I am making assumptions based on electronics 101 courses I took 15 years ago...

Either way I want to first make sure that I don't have a lemon so order of priorites is:

- Get a better PSU
- Apply best possible OC/UV with 100% power limit on GPU
- Install the n3rdware copper cooler
- Hope that the sheer cooler operation temps alone unlock a tiny bit more performance (NVidia often boosts higher at lower temps)
- Then see what's next.

The nice thing about the n3rdware cooler is that it might even allow VRM cooling with some thermal pads... I'll definitely look into that.
 
  • Like
Reactions: Arboreal

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,166
3,504
www.revoccases.com
Very interesting information, thanks!

My curve currently looks like this



But during the benchmark it never exceeds 1250 MHz or so... keep in mind I am running with a 97% power limit because my PSU is still causing occasional shutdowns at 100% (a beefier one is on the way from China...)

your curve looks fine to me, normally you should score about 6200 points setting it to 1600Mhz @ 0,750V (without shunt mod) so I assume your power limit is the culprit...

Speaking of the 12V cable, I also feel like it's a bit weird to just directly pipe an unfiltered 12V line from your PSU in parallel to 12V from the motherboard that traveled through a bunch of other components and probably has a very different looking ripple profile... I assume GPUs with proper PCIe power connectors don't do this the same way but also apply some filtering before combining all the 12V lines... but I am making assumptions based on electronics 101 courses I took 15 years ago...

the motherboard doesn't filter any ripple off the 12V PCIe rail, even if "properly" connected... on the GPU side, the additional 12V cable sits right at the shunt resistor, all other components/circuits come later... the only issue I see with this mod is the stress on the VRMs that are certainly not designed to handle 140W long term, at least not without additional cooling...
 

Snerual

Cable-Tie Ninja
Jul 3, 2020
196
224
Yeah actually after typing that response I realized what my issue was... I had both fan curve and thermal limit at stock... and apparently even if you come within 15C of thermal limit, the card already starts throttling just in case...

My best score with stock fan curve and case closed: 5997
Fan pegged at 100%, top of case removed: 6252
100% power limit (seemed to be fine in 3DMark at least): 6292

And then I got another idea... what if the memory is eating too much of the power budget? My VRAM is stable at +1500 (Matching the RTX 3060) but is this really needed considering the other limitations of this chip?

Reduce Memory OC from +1500 to +1250: 6328
Reduced Mem OC to +1000: 6330
Reduced Mem OC to +500: 6378

Important lesson learned: at these low TDPs, don't exaggerate on the memory OC. Memory also "steals" power budget! I assume this effect is even bigger on a 12GB A2000 like mine.

I still don't really understand why the GPU keeps downclocking during the run though... I would expect it to stay pegged at 1560MHz or so but by the end of Graphics Test 2 it is hovering around 1200 MHz.

Even better news: Also Superposition doesn't trip the PSU anymore at 100% power limit (and my score is up from 3300 ish to 3719). I assume the issue was transient spikes, and the flat freq/voltage curve completely irons those out (the chip can simply not go above 750mV anymore, ever.)
 

SFFMunkee

Buy first, justify later?
Gold Supporter
Jul 7, 2021
944
1,012
Yeah actually after typing that response I realized what my issue was... I had both fan curve and thermal limit at stock... and apparently even if you come within 15C of thermal limit, the card already starts throttling just in case...

My best score with stock fan curve and case closed: 5997
Fan pegged at 100%, top of case removed: 6252
100% power limit (seemed to be fine in 3DMark at least): 6292

And then I got another idea... what if the memory is eating too much of the power budget? My VRAM is stable at +1500 (Matching the RTX 3060) but is this really needed considering the other limitations of this chip?

Reduce Memory OC from +1500 to +1250: 6328
Reduced Mem OC to +1000: 6330
Reduced Mem OC to +500: 6378

Important lesson learned: at these low TDPs, don't exaggerate on the memory OC. Memory also "steals" power budget! I assume this effect is even bigger on a 12GB A2000 like mine.

I still don't really understand why the GPU keeps downclocking during the run though... I would expect it to stay pegged at 1560MHz or so but by the end of Graphics Test 2 it is hovering around 1200 MHz.

Even better news: Also Superposition doesn't trip the PSU anymore at 100% power limit (and my score is up from 3300 ish to 3719). I assume the issue was transient spikes, and the flat freq/voltage curve completely irons those out (the chip can simply not go above 750mV anymore, ever.)
I was about to chime in with very similar advice - power and temp limit are both crucial when you're aiming for those sweet sweet 3DMark graphics pts.

Yes, GDDR6 will definitely use a good chunk of your power budget, so you'll often want LOWER memory OC for a HIGHER core OC (and it's not even the same for Graphics1 and Graphics2! I even wonder if the FAN is considered within the power budget, so I don't set fan to 100%, usually opting for about 55-70% depending on how spicy the previous runs were (higher at higher voltages, 45% fan is plenty for 750mV runs).

FYI my A2000 can boost happily to a higher speed on test1 but crash in test2). Test 2, similar to Furmark will have a lower core clock than you'll see in Test 1.

You won't see it hitting the actual max frequency you select on your curve much, usually 200-300MHz below. From memory when I was selecting 1590MHz stock it was actually doing more like 1250MHz. I think maybe I led you astray there, I'd have to go back and look as I don't think I was doing 1650MHz curve when it was stock... Hmm.

By having the flat curve, you'll also eliminate those moments when the card tries to boost to some stupid frequency by shooting up to 1V or more, it would be nice if Afterburner could monitor the voltage too, but at least (IIRC) GPU-Z still can.

I did a whoooole lot of trial-and-error with my card, and suspect it's the only way to get the best out of each individual A2000 sample. I would typically follow this process (or reversed if trial/error testing voltage with a target frequency):
1) set power/temp limit at max with TEMP preference;
2) set fan at fixed point (e.g. 55%) - reduce if previous run stayed under 75C, increase if hitting 80C;
3) choose a target voltage (e.g. 893mV);
4) set it as the 'max' by dropping all V/F points after it
-- shift+click the next V/F point up,
-- shift+click the highest voltage point on the graph (highlighting
-- click+drag any of the selected points below your target voltage point;
-- click APPLY back in the main Afterburner window;
-- validate if the V/F curve shifts up/down a little bit by closing & reopening the curve window and checking your target points;
5) drag the whole curve up (alt + click/drag) to an estimated stable frequency;
-- click APPLY back in the main Afterburner window;
-- validate if the V/F curve shifts up/down a little bit by closing & reopening the curve window and checking your target points;
6) run your preferred test (I always use TimeSpy);

- if test completes, review the frequency range and temperature range
- if it crashed, consider higher voltage (only if shunt-modded) or lower frequency

Go back and start again. :)

GOOD LUCK!
-Nathan
 

Snerual

Cable-Tie Ninja
Jul 3, 2020
196
224
Since this is now the unofficial A2000 undervolting thread... I am now doing 3819 in Unigine Superposition and 6625 Graphics Score in Time Spy. (without mods or custom cooling)

This is with 1680 MHz at 743 mV. At 1700 MHz Superposition crashes so I will need to do some serious testing to make sure 1680 is truly stable. But overall I am not complaining! 10% performance increase over stock without any modding (in Superposition it's even more like 17% increase).

Edit: 1680 MHz is NOT stable in Assassin's Creed Unity... 1620 seemed OK but I need to do more extensive testing.
 
Last edited:

bingsoo

Chassis Packer
Feb 8, 2023
17
7
hmmm... for replicating I think you could (almost) get away without any special skills or equipment if you can find a local "maker space" or some online service for 3D printing and CNC machining. If you want to do something like that all DIY you'll need at least a 3D printer and maybe a hand tool to machine the heatsink. Here is a good example how the cutouts on the heatsink can be done without a CNC: Custom heatsink for the RTX A2000 | SFF.Network (smallformfactor.net)

if you want to do your own, similar mods there's a lot to learn. Like: CAD, soldering, 3D printing, CNC machining, etc...

as a starting point I would probably first try to get familiar with CAD (e.g. watch some YouTube videos and try designing some own parts). Afterwards I'd get a cheap 3D printer & a caliper and see where it goes from there...
Thanks @REVOCCASES! Appreciate your guidance!
 

b_force

Average Stuffer
May 28, 2019
73
28
Unfortunately not. I'm using a motherboard fan header and Rem0o/Fan Control which works great.

Only downside is that you have this small program running in the background but on the plus side, the A2000 pulls a little less off the PCIe slot (which might be good if I'm going to OC / shunt mod it?)



I've discussed this with Gunique / Gurywah long time ago in order to "cheat" some DELL Platinum Server PSUs, but even he didn't know of any solution... :\

Another idea for the A2000 was to make a 0-RPM fan mod adapter to get rid of the noise at least in idle, but overall I think Rem0o Fan control & motherboard fan header works best for me.

View attachment 2425
Since my rx6400 just died, I am reading through some alternatives.

If people are really interested, I could make some small pcb that can spoof the rpm signal.
But I think Fan Control is much more practical and useful anyway. (I run it on every pc I have).

Only on mini-ITX boards the amount of fan outputs can be very limited.
 

REVOCCASES

Shrink Ray Wielder
Original poster
REVOCCASES
Silver Supporter
Apr 2, 2020
2,166
3,504
www.revoccases.com
If people are really interested, I could make some small pcb that can spoof the rpm signal.

that would be awesome! @robbee and me were looking for a solution to this issue but couldn't find anything off-the-shelf...

I made this simple 0-RPM mod a while ago, using a thermo-switch but if you can find a more professional way...

1688608975319.png
 

robbee

King of Cable Management
n3rdware
Bronze Supporter
Sep 24, 2016
881
1,387
I did find some small programmable pcb's with a temperature sensor that could turn the temperature readout into a fan pwm signal. It could be useful for personal projects but I decided not to sell them with my heatsink as the quality was questionable and it adds a lot of complexity to the product.
 

b_force

Average Stuffer
May 28, 2019
73
28
that would be awesome! @robbee and me were looking for a solution to this issue but couldn't find anything off-the-shelf...

I made this simple 0-RPM mod a while ago, using a thermo-switch but if you can find a more professional way...

View attachment 2592
I can see what I can do (has more to do with time, I still have a day job you know 😉).

Just send me a PM with some details, requirements, size etc
 
  • Like
Reactions: BaK

b3rdm4n

Cable Smoosher
Feb 21, 2023
10
9
I was reading this thread and PM'd SFFMunkee about them and figured I should just come here to chat about it.

Really love my little A2000, in an Inwin Chopin build that's almost ready to post about too :) I've already done memory pads and repasted the card, I'm gathering parts for the shunt mod and also have a fan to PWM motherboard connector coming so I can power the fan from the motherboard and liberate another (apparently) 1-7w from the total board draw.

So to start I was going to go fairly mild with just 1 resistor and see how that goes, I'm comfortable soldering so I'll make up a positive wire to something 12v to supplement it. Not chasing many more watts just that little edge.
 

SFFMunkee

Buy first, justify later?
Gold Supporter
Jul 7, 2021
944
1,012
I was reading this thread and PM'd SFFMunkee about them and figured I should just come here to chat about it.

Really love my little A2000, in an Inwin Chopin build that's almost ready to post about too :) I've already done memory pads and repasted the card, I'm gathering parts for the shunt mod and also have a fan to PWM motherboard connector coming so I can power the fan from the motherboard and liberate another (apparently) 1-7w from the total board draw.

So to start I was going to go fairly mild with just 1 resistor and see how that goes, I'm comfortable soldering so I'll make up a positive wire to something 12v to supplement it. Not chasing many more watts just that little edge.
I found the single shunt mod didn’t do much, BUT I was gunning for fastest A2000 (purely because I could).

If you just want to be able to hold boost clocks a bit longer and slightly higher then you should be fine.

The consensus seems to be the frontside shunt is better/safer, if only doing the one. I’m not sure why YMMV

Happy to answer any further PMs or replies in here

Cheers
 
  • Like
Reactions: b3rdm4n

b3rdm4n

Cable Smoosher
Feb 21, 2023
10
9
I found the single shunt mod didn’t do much, BUT I was gunning for fastest A2000 (purely because I could).

If you just want to be able to hold boost clocks a bit longer and slightly higher then you should be fine.

The consensus seems to be the frontside shunt is better/safer, if only doing the one. I’m not sure why YMMV
At this point I'm thinking maybe a milder shunt then, and definitely the 12v extra wire.

Spent a lot of time over the weekend on the A2000 rig and it continues to impress me with it's chops at 1080p or even 1440p, tiny little beast! and for the time being the rest of the system is nothing special - i7 7700, 32GB DDR4 2400 and a Sata SSD. I think maybe in 12-18 months when I upgrade my main rig it'll end up with what's in that right now, which is a 5800X3D and 32GB @ 3600.
 
  • Like
Reactions: SFFMunkee