*Official Benchmark Thread* - Post it here or it didn't happen :D

[Rage Set]

On 2/17/2026 at 4:12 PM, Papusan said:

 

I think it will work fine but the vbios screw up the fans. It seems nothing is meant to work properly with Asus cards. I mean have seen the AIO owners say only the single fan for the VRM on the GPU  works but not for the fans for the radiator. Can probably swap out the org fans with some that I can connect to a fan controller/MB. The fans for the radiator coming with the Asus AIO cards doesn't have a "normal" fan connector or connected with cables. And not sure the fan for the VRM will work correctly anyhow. Everything proprietary mounted. Exactly as good old Dell. Our old friend F. Azor would  be proud of Asus and todays modern way to make tech.

With a custom block it will work just fine but that will be another $500-700 with needed parts. And +2 months lead time. Nothing to find for needed parts here home.

 

Jufus want you to pay for the repair, LOL

 

 

 

I am sorry, but the guy doesn't know what he's talking about. He claimed he was able to "warp" the internal PCB. Clearly, he hasn't watched Northwestrepair - the PCB's on these 5090's are literal heatsinks. He almost likely killed the GPU by a mistake in his shunt mod. 

 

This is what happens to people that labeled themselves experts. 

[electrosoft]

I started having weird little freezes with my main rig about a month ago. I caught the flu and was down for the count for 10 days and did not care about my computer, but getting back to D2D usage and just very random little issues. On the 17th the errors went into full tilt hardware errors overdrive:

 

 

That was when I pulled my 5090 to swap into my Ultra rig and it had no problems and my main rig "calmed down" a bit without the 5090 in there. Still, something was up.

 

Pulled out good 'ole AIDA full suite stress test, no problems.

 

Pulled out OCCT and ran extreme  tests on every sub system no problems.

 

Ran OCCT combined extreme and started getting random errors every different run within 5-60 seconds:  "VRAM failed," "GPU failed," "memory failed" and even "CPU failed". Each run was a crap shoot which error I would get so I was pretty sure it was the EVGA P2 1600w that was slowly failing and then suddenly took a turn for the worse.

 

Installed the MSI AI1600T tonight and combined passed with flying colors.

 

As a checksum, hooked the EVGA P2 1600w  to the Ultra Rig with the 9070xt and OCCT combined crashed and burned on it too  with the 9070xt. Bonus was seeing transients on the 9070xt hitting 600w+.....

 

Getting the AI1600T couldn't have come at a better time....

 

EVGA P2 1600w  has a 10 year warranty so it will be interesting to test the "new" EVGA for an RMA.....might as well get an RMA for a dead/dying K20 keyboard I have in the closet too from them. wasn't worth the postage to send it back but combined with the PSU probably worth the round trip.I tossed one of their wireless mice that just died a bit ago. Shoulda held onto it to toss in there too for an RMA......

 

Wife's EVGA K20 keyboard is still going strong and she beats the crap out of that thing playing WoW.....

 

MSI AI1600T is gorgeous.....too bad you'll never see it in this Phanteks case. Good thing is switching to a single 12v2x6 run and the way the cables are thinner and braided meant easier runs and management. It has a USB-C to USB connector to the MB, but you need to run MSI Center Bloatware  to access the features and yeah, no thank you. I'll use it as is.

 

 

Almost in final boss form ( need to make a few cuts to the brackets for the DDR5 fan):

 

 

-----------------------

 

On 2/19/2026 at 3:57 AM, jaybee83 said:

 

huh, nice lil side bonus if the WVPII decreases coil whine on the gpu 😄 u planning to test any of the 800+W vbios versions? or sticking to gigachad?

 

Yeah, that was a pleasant surprise on the Ultra when running the original MSI octopus connector directly vs through the WVP2. I'll take every bit of coil whine reduction I can get. 

 

Definitely going to go back to 800w Matrix as my daily vs just testing but I also want to test the 1000w Lightning. Nothing wrong with the gigachad, but I'm leaving a lot of the 1.125v cap on my card on the table with the 600w variants.

 

On 2/19/2026 at 11:18 AM, Mr. Fox said:

I normally do not because people like that are not worth the effort. Nothing said can address their mental illness. I started/stopped more than once before replying, but when I noticed him being an ass to multiple people I decided it was worth the risk. The response I posted was much kinder and more humorous than the few I started to post and did not. I also pinged the moderators to draw their attention to the butthead.

 

One thing I learned during the Pandemic was to scroll on by comments and positions that make zero sense. Better for your mental health. Politics? Religion? Absolutely stupid takes on tech? Scroll Scroll Scroll! I feel bad for family and friends that sit on forums and social media for hours on end just arguing. You won't walk away feeling better and you will carry that luggage with you for an indeterminate amount of time.

 

22 hours ago, Reciever said:

That was one of the main reasons I stopped surfing there. 

 

Idiots can run their mouths but if you engage you were treated the as committing the same crime. It was even worse with their vendor reps at the time. 

 

I hope it has improved over the years but with that site having sold it's IP several times over now the sense of identity it had is likely fizzled even moreso. 

 

Most forums and groups I quietly extract nuggets of useful information and absolutely refuse to engage in any type of flame wars. It's just not worth it.

 

I didn't join OCN till 4-5 years ago, so I have no idea what it was like during its heyday. 

 

21 hours ago, Talon said:

 

Are you on the latest BIOS? I updated last week and got around to tuning it a bit more. I've found on the latest BIOS, I can now tune VNNAON manually without issue, maybe I always could and never tried. I thought I remember it causing instability when touched around launch and I never tried again. Now latest BIOS, I can push D2D up to max 40x without issue, stable. I just can't use the Asus auto voltage for VNNAON. This sample needs 1.12v instead of auto 1.012v. I had a 285K that could do 40x auto voltage, but P cores were just ever so slightly worse so I kept this 38x-39x auto chip. Now I've been testing and using 40x for a couple days and it's having zero issues. Sure it doesn't amount to much, but more mental wanting to max it out lol. 

 

It's honestly a shame Intel has killed off 1851 entirely and 270K Plus will be all they release. My guess with same crap 40x limitations in place. Panther Lake has a very high D2D apparently and they've fixed the latency issues. 

 

I am not. I'll have to give it a whirl.

 

I'll have to check, but outside of outliers that heavily favor X3D (IE Fallout 76), Arrowlake is solid at 4k IMHO. I was a little bummed they cancelled the 290k. I was tempted to pick up that nice 285k over on the OCN forums that is delidded and 9200 C40 certified to play with....

 

Arrowlake felt a touch like Rocket Lake and almost a prototype calm before the big storm coming. I'm expecting big things at the end of the year from Intel.

 

13 hours ago, Papusan said:

 

It was just question of time. Asus couldn't be any worse than MSI. Hence this one come now. Sad Asus need to be pushed spanked by their competitors... Their reputation as THE PREFERED OC BRAND at stake. Asus own cards is almost none compatible for cross-flashing with any brands vbios versions (neither with firmware from their own SKUs). They really screwed up this gen. It's hard be on the list of shame. And Asus will stay there at least for this gen. Yep, their reputation is severely damaged. Shame on you. 

 

Another RTX 5090 2000W XOC BIOS leaks out, NVIDIA extreme OC scene is slowly waking up

 

ASTRAL RTX 5090 2000W XOC V2 leaked, its the newer version of the bios, which should include fixed power limit, with GPU able to reach 2000W now.

 

Gamers Ditch AMD Radeon RX 9000 GPUs Due To High Prices, Forcing Retailers To Drop The Prices

 

 

At least almost all of the XOC bios files are out in the wild now to play with so that's a plus....unless you're an FE owner or an Asus/Suprim AIO owner.

 

As for the 9070xt? Marketshare tells us if given a choice, gamers will pick Nvidia all day everyday still but with 5070ti prices through the roof, 9070xt is viable....if priced properly.

 

All the market conditions in the world won't force gamers to exceed a certain price point.

 

I knew the writing was on the wall when Newegg started lowering their 9000 series prices a few weeks ago. Either lower the prices or get stuck with a glut considering 9000 series has been on the market for almost a year and 5000 series for over a year. Many who wanted them have bought them.

 

Same theory applies to ram prices. You can jack up the prices all you want, but all I'm seeing is plenty of stock, eBay kits priced lower than the MSRP ones not moving and Newegg and MC now offering bundle deals.

 

Nevermind the price of many things overall has gone up in this economy. Comparing my grocery price history today vs a few years ago and it's a bloodbath. Comparing many items on subscription on Amazon and prices are up easily 10-15% overall if not more. Prices in restaurants is so outrageous now and/or shrinkflation we eat at home now or do pick up orders to avoid a savage 20%-23% "recommended" tip (remember when 15% was considered generous?)

 

Good times! 

[electrosoft]

11 minutes ago, Rage Set said:

 

 

I am sorry, but the guy doesn't know what he's talking about. He claimed he was able to "warp" the internal PCB. Clearly, he hasn't watched Northwestrepair - the PCB's on these 5090's are literal heatsinks. He almost likely killed the GPU by a mistake in his shunt mod. 

 

This is what happens to people that labeled themselves experts. 

 

He's fun to watch though..... 🤣

 

He definitely has a very small following but 40k isn't terrible and enough to grift a decent amount of $$$ his way via discord paywalls, tuning classes, custom tuning for $500 a pop and constantly soliciting donations, hawking his classes or directing his followers to publicly thank those who donate over and over again.....

 

 

[Rage Set]

3 minutes ago, electrosoft said:

 

He's fun to watch though..... 🤣

 

He definitely has a very small following but 40k isn't terrible and enough to grift a decent amount of $$$ his way via discord paywalls, tuning classes, custom tuning for $500 a pop and constantly soliciting donations, hawking his classes or directing his followers to publicly thank those who donate over and over again.....

 

 

 

I know soldering, PC break/fix and 3D printing extremely well, but I would never label myself an expert. If my YT channel ever reaches 50K+, you won't ever catch me trying to get money out of my community for a device that I damaged. I would probably be too embarrassed to even admit it, lol. 

[win32asmguy]

23 hours ago, Talon said:

Are you on the latest BIOS? I updated last week and got around to tuning it a bit more. I've found on the latest BIOS, I can now tune VNNAON manually without issue, maybe I always could and never tried. I thought I remember it causing instability when touched around launch and I never tried again. Now latest BIOS, I can push D2D up to max 40x without issue, stable. I just can't use the Asus auto voltage for VNNAON. This sample needs 1.12v instead of auto 1.012v. I had a 285K that could do 40x auto voltage, but P cores were just ever so slightly worse so I kept this 38x-39x auto chip. Now I've been testing and using 40x for a couple days and it's having zero issues. Sure it doesn't amount to much, but more mental wanting to max it out lol. 

 

It's honestly a shame Intel has killed off 1851 entirely and 270K Plus will be all they release. My guess with same crap 40x limitations in place. Panther Lake has a very high D2D apparently and they've fixed the latency issues.

 

Is that 285K that could do D2D 40x at auto voltage the one you sold me? I have not done much tuning with it sadly and will probably not end stretching it much. But it is still a major uplift compared to the 12900KS I used before especially with encoding jobs. It can even outperform the 24 P-core Xeon 2495X in certain tasks that could not multithread well beyond 8 P-cores. I sold my 5090 a few weeks ago as well as I still do most of my gaming on my laptop (which I am now testing out Prema's bios which has been really fun). I have even tested out my 32GB M-die kit which is XMP 6400 CL32. Obiously not as great as the 48GB M-die which would do 8400 CL40 out of the box. I need to bench my encoding jobs and see if the faster memory speeds them up. Then again if I want Nova Lake later this year the faster memory would probably pair better with it.

[Talon]

45 minutes ago, win32asmguy said:

 

Is that 285K that could do D2D 40x at auto voltage the one you sold me? I have not done much tuning with it sadly and will probably not end stretching it much. But it is still a major uplift compared to the 12900KS I used before especially with encoding jobs. It can even outperform the 24 P-core Xeon 2495X in certain tasks that could not multithread well beyond 8 P-cores. I sold my 5090 a few weeks ago as well as I still do most of my gaming on my laptop (which I am now testing out Prema's bios which has been really fun). I have even tested out my 32GB M-die kit which is XMP 6400 CL32. Obiously not as great as the 48GB M-die which would do 8400 CL40 out of the box. I need to bench my encoding jobs and see if the faster memory speeds them up. Then again if I want Nova Lake later this year the faster memory would probably pair better with it.

 

Yep! That 285K might have actually been better than my 285K I am running lol. I didn't really test the 2 fully, but that 285K I sold you did my 8400 CL38 tune, same voltages, ran 40x D2D no issue, and only had a very slightly higher P core voltage and ring voltage. But it still did the same 41x ring and 56x P cores. The chip I sold you also booted higher E core on auto voltages lol. I can boot the higher E core clocks on this chip, but the low voltage V/F stops it from scaling unless I go in adjust the V/F offset for point 8. I think that is mostly a function of interpolation failing, while your slightly higher voltage allows for better interpolation and boots with no adjustments. 

 

But ya my 285K can't do 40x D2D unless I run a manual voltage, but I've been gaming on it and testing it for a couple days now and no issue at 40x. If we had a higher ceiling you'd likely be able to go higher since you still have voltage room. 

[electrosoft]

6 hours ago, win32asmguy said:

 

Is that 285K that could do D2D 40x at auto voltage the one you sold me? I have not done much tuning with it sadly and will probably not end stretching it much. But it is still a major uplift compared to the 12900KS I used before especially with encoding jobs. It can even outperform the 24 P-core Xeon 2495X in certain tasks that could not multithread well beyond 8 P-cores. I sold my 5090 a few weeks ago as well as I still do most of my gaming on my laptop (which I am now testing out Prema's bios which has been really fun). I have even tested out my 32GB M-die kit which is XMP 6400 CL32. Obiously not as great as the 48GB M-die which would do 8400 CL40 out of the box. I need to bench my encoding jobs and see if the faster memory speeds them up. Then again if I want Nova Lake later this year the faster memory would probably pair better with it.

 

Whoa, you sold your 5090? I guess it makes sense considering you game at 1440p which keeps consistency across your desktop and laptop 1440p displays. In WoW, so much performance is left on the table at 1440p with a 5090. My wife games at 1440p and the 9070xt is near perfect for that at Ultra RT Max, but if given a choice I'd go with a 5070ti depending on price. Hope you got a good price for it.

 

Did you ever tune and push those V-color 9600 sticks? Those should slot in nicely for Nova....

 

You're the second to sell off their primary/only 5090 so far. You and @tps3443. I wonder if he kept his Acer laptop or sold that too?

 

 

 

 

 

 

[Mr. Fox]

For the past 6+ years I have used my own desktop PCs for work and now I will be using a company-issued turdbook for work. I repurposed my dual Acer 4K 160Hz panels for the work PC since I will be needing the massive screen real estate for work. I grabbed a 40" Samsung Odyssey G75F 5120x2160 180Hz WUHD monster on sale at Micro Center for $638 (model LS40FG75DENXZA) and it seems pretty great so far. It doesn't like the 4090 and neither does my ASUS 4K 120Hz monitor. I have to lower the resolution to not have a BIOS or Windows Boot Manager black screen (same as the ASUS screen) when using DisplayPort. Something about 40-series firmware and high refresh rate 4K on DisplayPort is glitchy. There are many examples of people complaining about this online. Using HDMI 2.1 it functions flawlessly. The 5090 doesn't care whether I use DisplayPort or HDMI.

 

 

[Reciever]

I bought another 16x to x8x8 adapter that is a bit more direct and doesnt use the SAS interconnects + daughterboards.

 

If this doesnt work then I'll just have to eat the L and buy the adapter from C_Payne as he had tested his adapter with my motherboard years ago as working. Trouble is the guy is not based in the US so i'll have to risk the overseas + customs. Really want to test 7900 XTX and 3090Ti together though I could go with less I am enabling the idea of getting a second 1080mm UT45 for "symmetry" purposes under the AC unit. If so then I would likely go for a waterblock on the 5800X3D just to make it easier on my fingers.

Also bought another PSU for the benching rig. Be Quiet 1000w. If I need more power I still have 3x 900w for GPU's.

Going to be changing up my DD a bit as the current scenario is not really optimal for benching. God of War: Ragnarok took the GPU to 39c the other night so I think I am too close to the wall and recirculating air.