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

[Mr. Fox]

20 hours ago, Mr. Fox said:

Definitely power-limited. There is no point in repeating these benchmarks without a shunt mod because it's maxed out on the power budget now.

I only need two, but I will have spares. I am going to try the Jufes no detection easy reversal sneaky technique this time instead of solder and see how that works. If it doesn't I can always solder it after the fact. If we eventually get a 1kW XOC vBIOS leaked then I can just peel them off without a trace and without any effect on the warranty. If it works well I will send you a couple for your 5090 @electrosoft

 

https://www.mouser.com/ProductDetail/755-PMR100HZPFV2L00

 

[electrosoft]

So strange a few days ago to see 5090FE's available on Best Buy almost all day with the, "See Details" button blue before they eventually sold out. I didn't want one (explained below), but still good to see. Makes sense considering in the EU they are actually being discounted now a bit in certain regions.

 

------

 

I'm definitely locked onto the Jonsbo Z20 MATX portable case now after seeing it can fit even an Astral 5090 in it:

 

Check out the temps which are great too.

 

 

Astral has almost the exact same dimensions as a Vanguard 5090. This case was in my top 3 before but after seeing it can fit a Vanguard 5090 sized card and actually seeing it in action here at Quake Con from a gent in the BYOB line, I'm sold on it. It seems damn near perfect and checks all the boxes..... plus it's only $99.99 now. I especially like the no riser design.

 

 

Timestamped with gent in line.....

 

It can handle monster quad slot behemoths like the Astral, Vanguard and more. Looks like a fun case to play around with too.

 

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

20 hours ago, Mr. Fox said:

I figured I had better do it while I had the wild hare up my butt or I'd end up deciding it wasn't worth the effort. 🤣 I wheeled everything into the living room and benched from the comfort of my recliner.

 

Definitely power-limited. There is no point in repeating these benchmarks without a shunt mod because it's maxed out on the power budget now. But the clocks are holding a lot higher with the chilled water. These were pulling less than 1.000V under load. +300 core offset. I will have to order the resistors Brother @tps3443 mentioned since all of my spares are too high resistance. Travis do you have a link where you bought yours? Digi-Key or Mouser?

Speedway - https://www.3dmark.com/sw/2656446 | https://hwbot.org/benchmarks/3dmark_-_speed_way/submissions/5888556
 Steel Nomad (DX12) - https://www.3dmark.com/sn/8377250 | https://hwbot.org/benchmarks/3dmark_-_steel_nomad_dx12/submissions/5888557

Steel Nomad (Vulkan) - https://www.3dmark.com/sn/8377374 | https://hwbot.org/benchmarks/3dmark_-_steel_nomad_vulkan/submissions/5888558?

Port Royal - https://www.3dmark.com/pr/3623727 | https://hwbot.org/benchmarks/3dmark_-_port_royal/submissions/5888560?

Solar Bay - https://www.3dmark.com/sb/367617 | https://hwbot.org/benchmarks/3dmark_-_solar_bay/submissions/5888561?

Solar Bay Extreme - http://www.3dmark.com/sb/367638 | https://hwbot.org/benchmarks/3dmark_-_solar_bay_extreme/submissions/5888562

Time Spy - https://www.3dmark.com/spy/58438069 | https://hwbot.org/benchmarks/3dmark_-_time_spy/submissions/5888565

Time Spy Extreme - https://www.3dmark.com/spy/58438160 | https://hwbot.org/benchmarks/3dmark_-_time_spy_extreme/submissions/5888567

 

8 hours ago, Mr. Fox said:

https://hwbot.org/benchmarks/cinebench_-_r23_multi_core_with_benchmate/submissions/5888802

 

4 hours ago, Mr. Fox said:

OK I think I am done now

 

Fire Strike - https://www.3dmark.com/fs/33747306 | https://hwbot.org/benchmarks/3dmark_-_fire_strike/submissions/5888849
 
Fire Strike Extreme - https://www.3dmark.com/fs/33747317 | https://hwbot.org/benchmarks/3dmark_-_fire_strike_extreme/submissions/5888851

Fire Strike Ultra - https://www.3dmark.com/fs/33747324 | https://hwbot.org/benchmarks/3dmark_-_fire_strike_ultra/submissions/5888853

 

 

Chiller mode ACTIVATED...

 

The Mrs. was ok with a living room benching session? 😁 My wife gives me deliberate side eye when I start to use the dining room table for projects.....

 

45 minutes ago, Mr. Fox said:

I only need two, but I will have spares. I am going to try the Jufes no detection easy reversal sneaky technique this time instead of solder and see how that works. If it doesn't I can always solder it after the fact. If we eventually get a 1kW XOC vBIOS leaked then I can just peel them off without a trace and without any effect on the warranty. If it works well I will send you a couple for your 5090 @electrosoft

 

https://www.mouser.com/ProductDetail/755-PMR100HZPFV2L00

 

 

Looking real good! Once that shunt is in place that Zotac is going to really start to sing full volume....

 

lol, me and shunting.... 🤣 Right after @Papusan and @Talon shunt theirs first. 🙂

 

 

14 hours ago, Papusan said:

 

This what you said while Intel almost say Core Ultra (Arrow Lake) was a failure as desktop chips 😎 You and Intel's CFO can't both have correct brother, LOL

 

One of you has to be wrong😂

 

As you know, we kind of fumbled the football on the desktop side, particularly high performance desktop side. So we’re as you kind of look at share on a dollar basis versus a unit basis, we don’t perform as well and it’s mostly because of this high end desktop business that we didn’t have a good offering this year.

 

Intel Confirms Arrow Lake Weakness, Bets on Nova Lake to Compete With AMD Zen 6

 

At the Deutsche Bank Technology Conference in 2025, Intel’s CFO David Zinsner explained that the company sees Arrow Lake as a step that didn’t quite hit the performance target for high-end desktops.

 

Well, he clearly is talking the general (m)asses that didn't take the time to properly tune it.

 

It is 13th vs 7800X3D all over again. Everybody screaming X3D superiority among the general end users and channels, and when I decided to tune them both up, guess what? 13900KS made massive performance improvements and the 7800X3D marginally so since its main strength lies in that cache.

 

I'm not saying an Ultra 285k is going to slay a 9800X3D, but it can definitely gain a lot of ground when properly tuned. I'm just fighting the urge to do it all over again like before because that itch is picking up..... 

[Mr. Fox]

1 hour ago, electrosoft said:

Chiller mode ACTIVATED...

 

The Mrs. was ok with a living room benching session? 😁 My wife gives me deliberate side eye when I start to use the dining room table for projects.....

No she would not be. But, she is in WA visiting relatives and friends. I have 2 more weeks of having free reign of the house and then I will have to retreat to my hot little laboratory.

 

The dining table seems to be a universal taboo with the ladies. If I scratch it that is the end of the world. If the grandkids scratch it, that is adding "character" to the table, LoL. 🤣

[Mr. Fox]

Ooops. No chiller on this one. Voltage got higherwith the comparatively elevated temperature.

https://hwbot.org/benchmarks/cinebench_-_r23_multi_core_with_benchmate/submissions/5889007

[Papusan]

 

Hmmm. What can go wrong if you go cheapo?...

 

The documented case on Reddit should therefore not be viewed as an isolated incident, but rather as the direct consequence of a design weakness that is likely to be systematically present throughout the series.

 

https://www.igorslab.de/ursachenforschung-kondensatordefekt-auf-einer-pny-rtx-5070-mit-ansage-und-vorheriger-warnung/

 

Then we have the much more expensive 5090s from PNY goes booooom. Good components doesn't mean a shit if the design isn't up to par. Asus is greedy bastards but they can eletrical design.... Just too expensive for what you get. Same with Galax/KFA2's better cards.

https://www.guru3d.com/story/geforce-rtx-5090-goes-boom-pny-gpu-suffers-capacitor-failure-near-16pin-connector/

 

The Intel.... i mean the AMD moment. Also why you see Nvidia is so restricted with their boost throttling algo at awful low temperature. The silicon can't handle die shrink with reasonable voltage. Throttle the boost clocks below 40C will be the next for processors. Chips that handle 82-100C is for the older days. 

 

https://videocardz.com/newz/same-user-loses-another-ryzen-7-9800x3d-on-asrock-motherboard

[Reciever]

On 8/26/2025 at 4:39 AM, Mr. Fox said:

Congratulations on the promotion(s) brother. I have no doubt they were well-deserved.

 

Honestly, that 1080 Nova 360x360 radiator and case are as good or better than a MO-RA 360. Works just as well, costs less, weighs a whole lot less and takes up less space than a MO-RA, just not as flashy-looking. I loved mine and sometimes wish I had not sold it. @tps3443 had excellent service from his, too. If I were still using an open bench that would be the one I recommend. You can mount your D5 pumps/reservoirs directly to each side of it. As far as fans go, you need very high static pressure. Arctic P12 MAX or P12 Pro are the best bang for the buck. Don't spend more for something not better that makes more noise.

 

Here is a parts list I recommend for the external radiator build based on what I have used.

 

High Static Pressure Fans:

https://www.amazon.com/ARCTIC-P12-Pro-PST-CO/dp/B0DJDCBCC2

-or- https://www.amazon.com/dp/B0DJDBV48H

Reasonably priced, high quality and fantastic 6.9 mmH₂O static pressure. The outer ring on the tips of the fan blades is a design advantage and makes the blades stronger as well as being more effective.

 

Power Supply:

https://www.amazon.com/dp/B0CYZMK89X?th=1

You can buy molex or SATA extensions (depending on what your pumps use) like these and cut the input end off to create two pigtail connectors to solder the output ends to the power and ground wires on the PSU to run the fans and pumps. This type of PSU unit is compact and looks better than a bunch of extra un-used wires. Before I discovered this type of PSU I had used these before as well, which worked fine but were a lot more bulky in comparison. 

 

Fan Hub and Speed Control:

https://www.amazon.com/dp/B0DS2273B8?th=1

I recently discovered this item and besides working well for fan speed it is more affordable than what I used in the past (a Noctua control with a manual knob that costs a lot more) that required a separate fan hub. This will run all 9 fans with SATA power from the PSU.

 

Custom Fan Filter:

To help keep the fans and radiator clean and free of lint and dust, I made a custom filter using a magnetic black fiberglass window screen similar to this cut-to-fit kit. Easy to remove and clean and looks like it was made specifically for the application when all is said and done. The steel Nova 1080 radiator case supports the magnetic attachment.

 

D5 Pump and Reservoir:

I have owned two of these and they are excellent and affordable. They fit well on the sides of the Alphacool 1080 Nova radiator housing. Alphacool Core 100 Aurora reservoir Acetal/Acryl with VPP Pump. You can attach the black acetal pump block directly to the radiator case using the included machine screws. No need for the included brackets. Just drill the holes in the radiator housing and bolt it on. Tip: use the bracket designed to bolt the pump body to a 120/140mm fan as a template for drilling the holes to bolt it to the case. No need to measure the distance between the holes on the pump body before drilling.

 

Perfect, wasnt expecting a full write up but it is very much appreciated. I will probably start collecting more things towards the end of the year as I would like to get some benchmarking in the Winter when I can take advantage of the colder weather with some vacation time I think I still have a couple of weeks worth of time I can use. 

For the time being though, I did pick up 2x 9800 GTX+'s (55nm) to go with my accellero xtreme 9800's if the old posts are to be believed they dropped temps by around 30C+ so adding some cold air should let me around any temperature concerns for the most part. 

I also picked up a R9 390X 8GB TRI-X for 40 USD. 

If anyone here happens across some aftermarket heatsinks please feel free to reach out. I am hoping to collect enough for the GPU's I have, some of them are getting quite rare.

I guess at this point starting to collect Waterblocks can be the new hobby once I start diving into that space.

[Papusan]

 

Bro @Mr. Foxnext purchase. A pukebox friom one of the best known hw manufacturers. Also very known KB manufacturers. This to clean your stomach. Empty everything. Not only a Razer problem. More the normal from modern technology. $4000 down the drain. Even $50-100 is too much for this junk.

 

 

 

[electrosoft]

Well, this tracks with Steam Survey and AMD's new cards basically having no break through while NVIDIA continues to grow. They now own 94% of the market up from 88% last year this time. Insane....

 

but it makes sense considering how closely the 9070xt has been priced to the 5070ti. $599 vs $749 and the 9070xt makes sense. Any other pricing and it's a no brainer for the most part to go with a 5070ti.

 

Other tiers are just too closely priced to make an AMD purchase a potential no brainer. AMD needs to compete on price and performance.

 

Understand, for every 100 GPUs sold, 94 go to Nvidia and 6 go to AMD. I'd like to see numbers dialed down explicitly to gaming but those numbers are bonkers.

 

Newest Steam survey released just yesterday shows the 5060 is the fastest growing GPU and yet again AMD has not made enough of an impact to enter the charts.....

 

https://videocardz.com/newz/jpr-nvidia-discrete-gpu-market-share-reaches-94

 

 

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

 

Insane gains, basically same price $299.99 (even cheaper with inflation calculations)....many will still complain because it's their nature....

 

 

[Papusan]

5 hours ago, electrosoft said:

Well, this tracks with Steam Survey and AMD's new cards basically having no break through while NVIDIA continues to grow. They now own 94% of the market up from 88% last year this time. Insane....

 

but it makes sense considering how closely the 9070xt has been priced to the 5070ti. $599 vs $749 and the 9070xt makes sense. Any other pricing and it's a no brainer for the most part to go with a 5070ti.

 

Other tiers are just too closely priced to make an AMD purchase a potential no brainer. AMD needs to compete on price and performance.

 

Understand, for every 100 GPUs sold, 94 go to Nvidia and 6 go to AMD. I'd like to see numbers dialed down explicitly to gaming but those numbers are bonkers.

 

Newest Steam survey released just yesterday shows the 5060 is the fastest growing GPU and yet again AMD has not made enough of an impact to enter the charts.....

 

https://videocardz.com/newz/jpr-nvidia-discrete-gpu-market-share-reaches-94

 

 

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

 

Insane gains, basically same price $299.99 (even cheaper with inflation calculations)....many will still complain because it's their nature....

 

 

 

Nvidia have sold more 5000 series cards than anyone though at the beginning of the year. Another nail in the AMD cards coffin. Right into the 9070XT heart. And they need to stop price their products so close to Nvidia. Won't work. Fire Azor!!! Today.

 

Rumor: Nvidia to Replace RTX 5070 Ti and 5080 With SUPER Models in October 2025

Published 13 hours ago by Hilbert Hagedoorn

 

If true, this move suggests Nvidia wants to lock down the market for graphics cards priced above $550 while future-proofing them with larger memory pools.

[Mr. Fox]

1 hour ago, Papusan said:

If true, this move suggests Nvidia wants to lock down the market for graphics cards priced above $550 while future-proofing them with larger memory pools.

Too bad for the dumb-dumbs that already wasted money on the latest GPUs with only 8GB. Sucks to them, but also not cool on NVIDIA's part to screw them over like that.

 

For anyone that believes NVIDIA actually wants to futureproof (i.e. sell fewer) GPUs, I have some ocean beach property in Phoenix that I am selling very cheap.

 

[Mr. Fox]

Alienware has met its match... they no longer the hold the title to "worst" thanks to HP. They are now the runner up.

 

[electrosoft]

2 hours ago, Papusan said:

 

Nvidia have sold more 5000 series cards than anyone though at the beginning of the year. Another nail in the AMD cards coffin. Right into the 9070XT heart. And they need to stop price their products so close to Nvidia. Won't work. Fire Azor!!! Today.

 

Rumor: Nvidia to Replace RTX 5070 Ti and 5080 With SUPER Models in October 2025

Published 13 hours ago by Hilbert Hagedoorn

 

If true, this move suggests Nvidia wants to lock down the market for graphics cards priced above $550 while future-proofing them with larger memory pools.

 

The scary part is Q1, ok, I get it since AMD had not released the 9070xt and Nvidia was the only new game in town so 12%->8%, but Q2 when AMD was fully on the market brand new and Nvidia's market share grew yet another 2% while AMD contracted that same 2%.

 

AMD has to get it through their thick skull that they will need to be better and priced super competitive to wrestle away customers because if not, why would they entertain switching? I mean, I held onto my 9070xt because it is an absolute 1440p WoW beast with the newest drivers and WoW updates. I am blown away at its performance and smoothness on the wife's 32" 165hz Asus display.

 

Even back in the 5090 camp, I still stand by the 9070xt handling the lows better than the 5080 and 5090 in WoW. That hasn't changed.

 

But those prices are going to continue to hold AMD back.

 

 

 

 

[Mr. Fox]

4 hours ago, electrosoft said:

 

The scary part is Q1, ok, I get it since AMD had not released the 9070xt and Nvidia was the only new game in town so 12%->8%, but Q2 when AMD was fully on the market brand new and Nvidia's market share grew yet another 2% while AMD contracted that same 2%.

 

AMD has to get it through their thick skull that they will need to be better and priced super competitive to wrestle away customers because if not, why would they entertain switching? I mean, I held onto my 9070xt because it is an absolute 1440p WoW beast with the newest drivers and WoW updates. I am blown away at its performance and smoothness on the wife's 32" 165hz Asus display.

 

Even back in the 5090 camp, I still stand by the 9070xt handling the lows better than the 5080 and 5090 in WoW. That hasn't changed.

 

But those prices are going to continue to hold AMD back.

NVIDIA would have to really eff things up or price themselves out of business (which they almost did) to get most people to move to AMD. Stupidity and screw-ups  on an order of magnitude and as unpardonable as removing hyper-threading and focusing on smartphone cores are the only reason my house is no longer an AMD-free zone. AMD did not take the win as much as Intel losing focus on what matters and giving it to them. All three companies suck and are run by idiots that suck at their jobs. Actually, that is the new normal for the industry as a whole.

[Papusan]

 

And it won't be any better as long they cannot do it right. You shouldn't need to have latest MB to make it work with newest gen graphics cards. 

 

Not being able to install a new generation graphics card is always a shame, even if the platform is starting to get old.

 

RX 9070 and X99 motherboard: vBIOS incompatibility

 

https://overclocking.com/rx-9070-et-carte-mere-x99-une-incompatibilite-au-niveau-du-vbios/

 

 

And here is another failure from big US tech companies. Will US adm now start subsidize asia'n tech companies? Let Intel feed the *money they got for free* back to their competitors pockets? US tax payers being heavily screwed as it seems. Intel's best new node from own US chips foundry only usable for making their iGPU tiles. What a joke. Or better say tragic. And this years to come. I really hope China don't go into war and take Taiwan. Because Intel have nothing to offer if China grab TSMC.

 

This reinforces the fact that Intel 18A-P could end up in Nova Lake-S like the Tile GPU, and not the Tile CPU , which would arrive in TSMC's N2P , or at least is on track

 

Intel admits TSMC's superiority: "We're going to use them to manufacture our products forever; they're a great partner."

Intel has been trying to regain ground in chip manufacturing with IFS for years, but the reality is that it still relies on TSMC to produce some of its most important products. So sad.

[Papusan]

7 hours ago, Mr. Fox said:

NVIDIA would have to really eff things up or price themselves out of business (which they almost did) to get most people to move to AMD. Stupidity

 

As long our old friend F. Azor stay in the Red corners gaming divisjon leadership, Nvidia have nothing to be afraid of. E-sport Games he said don't need AMD cards with more than 8GB vram, LOL

 

He even believe it himself. Or maybe the greedy mind come first. 

 

Stupids. 

[Mr. Fox]

5 minutes ago, Papusan said:

 

As long our old friend F. Azor stay in the Red corners gaming divisjon leadership, Nvidia have nothing to be afraid of. E-sport Games he said don't need AMD cards with more than 8GB vram, LOL

 

He even believe it himself. Or maybe the greedy mind come first. 

 

Stupids. 

We are surrounded by stupid.

Here is some new well-deserved Razer hate... https://youtu.be/8w8m1UuLsEQ

 

I received the shunt resistors yesterday so I will probably do the mod tomorroww.

[electrosoft]

20 hours ago, Papusan said:

 

Nvidia have sold more 5000 series cards than anyone though at the beginning of the year. Another nail in the AMD cards coffin. Right into the 9070XT heart. And they need to stop price their products so close to Nvidia. Won't work. Fire Azor!!! Today.

 

Rumor: Nvidia to Replace RTX 5070 Ti and 5080 With SUPER Models in October 2025

Published 13 hours ago by Hilbert Hagedoorn

 

If true, this move suggests Nvidia wants to lock down the market for graphics cards priced above $550 while future-proofing them with larger memory pools.

 

Nvidia doubling down on AMD and at least finally admitting gamers want more vram. 18gb and 24gb variants along with the 5070 getting a performance boost. It will close the gap between the 5070 and 9070 and give the 5070 more vram.

 

24gb variants of the 5070ti and 5080 just put the nail in the coffin for ther 9070xt atm unless prices seriously change.

 

19 hours ago, Mr. Fox said:

Too bad for the dumb-dumbs that already wasted money on the latest GPUs with only 8GB. Sucks to them, but also not cool on NVIDIA's part to screw them over like that.

 

For anyone that believes NVIDIA actually wants to futureproof (i.e. sell fewer) GPUs, I have some ocean beach property in Phoenix that I am selling very cheap.

 

 

5060 8GB makes sense for budget minded buyers. 5050 is a farce. At least AMD and Nvidia give buyers a choice between 8GB and 16GB models of the 9060 and 5060ti. Recent market data shows buyers are greatly gravitating towards the 16GB variants....shocker.

 

Yeah, Nvidia wants to lock down the market even more. We need AMD. They keep Nvidia on their toes at least a little bit. Only reason we're getting Supers with memory bumps along with maybe a bit of stagnant sales atm from those who want more vram.

 

Future proof? I'm sure that is blasphemy in the halls of Nvidia....

 

18 hours ago, Mr. Fox said:

Alienware has met its match... they no longer the hold the title to "worst" thanks to HP. They are now the runner up.

 

 

"No one can out screw up a known boutique mash up like Dell and Alienware!!!"

**HP and Voodoo have entered the chat**

 

🤣

 

2 hours ago, Mr. Fox said:

We are surrounded by stupid.

Here is some new well-deserved Razer hate... https://youtu.be/8w8m1UuLsEQ

 

I received the shunt resistors yesterday so I will probably do the mod tomorroww.

 

w00t! Looking forward to it! Seeing results from stock, AC, water, chilled and now shunted shows a nice progression. 

 

 

[Papusan]

13 minutes ago, electrosoft said:

5060 8GB makes sense for budget minded buyers. 5050 is a farce. At least AMD and Nvidia give buyers a choice between 8GB and 16GB models of the 9060 and 5060ti. Recent market data shows buyers are greatly gravitating towards the 16GB variants....shocker.

 

5050 should be the only card this gen Nvidia with 8GB vram. And 5060 with 16GB vram shouldn't be much more expensive than the 8GB variant. Maybe +25$ on top at best. 

[electrosoft]

6 minutes ago, Papusan said:

 

5050 should be the only card this gen Nvidia with 8GB vram. And 5060 with 16GB vram shouldn't be much more expensive than the 8GB variant. Maybe +25$ on top at best. 

 

In an ideal world, absolutely but Nvidia will eventually be brought kicking and screaming into more baseline vram both mobile and desktop from top to bottom.

 

With that being said, are there actually any games that can't run on 1080p medium with 8GB of vram?

 

I do think Nvidia shot themselves in the foot when they offered the 3060 with 12GB. They opened Pandora's box a smidge... 🤣

 

If Nvidia is moving the baseline with the 070 class on up now with Supers, and slowing orders of 8GB 060ti variants, eventually it would make sense to:

 

050  = 8GB

060  = 12GB

060ti =16GB

070   =18GB

070ti =24GB

080   =24GB

090    =32GB

 

The only fear of course is when 6000 series drops they regress back to non Super configs.

[Papusan]

2 hours ago, electrosoft said:

 

In an ideal world, absolutely but Nvidia will eventually be brought kicking and screaming into more baseline vram both mobile and desktop from top to bottom.

 

With that being said, are there actually any games that can't run on 1080p medium with 8GB of vram?

 

I do think Nvidia shot themselves in the foot when they offered the 3060 with 12GB. They opened Pandora's box a smidge... 🤣

 

If Nvidia is moving the baseline with the 070 class on up now with Supers, and slowing orders of 8GB 060ti variants, eventually it would make sense to:

 

050  = 8GB

060  = 12GB

060ti =16GB

070   =18GB

070ti =24GB

080   =24GB

090    =32GB

 

The only fear of course is when 6000 series drops they regress back to non Super configs.

 

I doubt we will see 8GB xx60 cards with 6000 series. That even without the Super moniker. Then they will differenciate with TDP and boost clocks for Super vs vanilla. For xx90 I don't know if they will try with 48GB vram. This probably depends on the bus/bits. Or 48GB to please creators. For a price hike as with 5090 above 4090.

[Mr. Fox]

4 hours ago, electrosoft said:

w00t! Looking forward to it! Seeing results from stock, AC, water, chilled and now shunted shows a nice progression. 

Mission accomplished. 100% working. My UPS is screaming bloody murder now and a quick MSI Kombustor run is now showing nearly 1200W getting pulled from the wall, whereas before it was between 850-900W doing the same thing.

 

It was not possible to do the Jufes incognito shunt mod because this new style of resistor has the solder points lower than the black plastic part of the resistor. Older resistors were the opposite, with the solder points on each end being flush or taller than the plastic. The resistors don't make contact as the solder points unless you use solder to bridge the air gap with this newer design. The dimensions are also smaller, so even if I had the right resistors in terms of resistance they would have been too long to work. So, she is soldered. I figured out on the second resistor that it worked easier to flip it over with the labeled side facing the original resistor. It was easier to hold in place for soldering with them face-to-face instead of labeled side up due to the shape of them.

 

I have to leave for a doctor's appointment in a few minutes but I will do more testing tonight and this weekend. Will most likely hook up the chiller tomorrow. Here are the PCB photos. Very well-made GPU.

 

Shunted resistors...

Looks like Kombustor was drawing around 800-900W. @tps3443

[Mr. Fox]

So, as best I can tell just playing a game like DOOM: The Dark Ages with no overclock the GPU is pulling around 700W now and between 950 and 1000W from the wall now. Boost clock is a little over 3000 MHz and no fluctuation.... just a flat line. Speedway the same. Stock boost is a flat-line 3000 MHz and about 1100W from the wall. runing Speedway. The power cable is not even getting a little bit warm to the touch.

 

I ordered a new Cyberpower 1500W UPS because both of the 800W untis that I have power off under 100% 3D load. They used to beep once in a while using the 4090 but never powered off. They clearly cannot handle what this shunted 5090 is requiring of them.

 

I also ordered a new WireView Pro with the down-facing connector so I can actually see how many watts are getting pulled through the 12VHPWR cable. GPU-Z is showing like 480W paying DOOM, so it may be more than 700W just stock. Both of those will be here tomorrow.

 

Here's the stock clock flat-line graph from Speedway. Before the shunt mod a stock run was something like 2850-2870 boost clock and constantly fluctuating. So, without touching anything is running much more aggressively now.While I am waiting for those items to arrive I am going to switch to the stock vBIOS and see if the behavior is the same, better or worse compared to the GigaChad vBIOS after the shunt mod.

[win32asmguy]

On 8/21/2025 at 11:53 AM, electrosoft said:

Yeah, the A18 was an absolute beast wrecking this 275HX+5070ti and I know most don't care about the fan noise but it was the ultimate nail in the coffin. Temps were really good across the board but those fans..... especially using the Alienware 2025 18 variant. Lights out temps, fantastic noise levels and even setting both to overdrive modes for max fans, the Alienware had not only better dB levels but better pitch. The Alienware defaults to solid teal now and the overhaul on the design is nice.

 

Started testing out the AW 18.

 

It came with Sunon fans that are not whiny which is nice. I agree the pitch is improved and the noise levels in general. I can also have this setting on my laptop stand without it causing whine or a pitch change which is much better than M18 R1. Its definitely a better air cooling solution than the Hydroc G2.

 

The RGB was Aurora by default and overriding to teal still does not set it as a boot effect. Ideally I want no RGB on any subzone by default (may have to unplug the mobo connectors) and white only on the keyboard. I could even live with solid teal everywhere but it seems like the RGB controllers don't have the few KB needed to store a profile to memory...

 

Installed the Kingston Fury XMP 6400 kit. It booted at 4800 but after enabling XMP it was not booting at all. I have heard that SREP still works on these so maybe manual tuning is possible. It will be painful though as not even a RTC reset is recovering from these memory brick scenarios. There is supposed to be a validated ADATA 7200MT 2x16GB XMP kit coming for these but I cannot find any info about it other than a part number in a support document.

 

Its a shame these still ship with E31. This is a perfect example why I think they should ditch it and use PTM7958SP, just too much risk of it leaking through the barrier. I would certainly be fully removing the board and carefully replacing it before it became an issue. I still hear reports of wide temperature variance on these new models which likely is from poor QC with E31 application.

 

 

[tps3443]

10 hours ago, Mr. Fox said:

So, as best I can tell just playing a game like DOOM: The Dark Ages with no overclock the GPU is pulling around 700W now and between 950 and 1000W from the wall now. Boost clock is a little over 3000 MHz and no fluctuation.... just a flat line. Speedway the same. Stock boost is a flat-line 3000 MHz and about 1100W from the wall. runing Speedway. The power cable is not even getting a little bit warm to the touch.

 

I ordered a new Cyberpower 1500W UPS because both of the 800W untis that I have power off under 100% 3D load. They used to beep once in a while using the 4090 but never powered off. They clearly cannot handle what this shunted 5090 is requiring of them.

 

I also ordered a new WireView Pro with the down-facing connector so I can actually see how many watts are getting pulled through the 12VHPWR cable. GPU-Z is showing like 480W paying DOOM, so it may be more than 700W just stock. Both of those will be here tomorrow.

 

Here's the stock clock flat-line graph from Speedway. Before the shunt mod a stock run was something like 2850-2870 boost clock and constantly fluctuating. So, without touching anything is running much more aggressively now.While I am waiting for those items to arrive I am going to switch to the stock vBIOS and see if the behavior is the same, better or worse compared to the GigaChad vBIOS after the shunt mod.

I’m glad it’s working well! Also I wanted to tell you I would not use a wire view pro. I know they are nice and very useful. But that’s primarily for a stock 5090 pulling no more than a sustained 600 watts. I’ve seen some issues of cables so hot using the wire view pro on shunted 5090’s, that melting is imminent. I would run just a straight cable to your 5090. Return the wire view pro. I’m totally blind on my cable. I had a sensor on it, but I trust it now. It’s been great and that’s inside a closed case, so I let the sensor monitor my case temp instead.

[Mr. Fox]

42 minutes ago, tps3443 said:

I’m glad it’s working well! Also I wanted to tell you I would not use a wire view pro. I know they are nice and very useful. But that’s primarily for a stock 5090 pulling no more than a sustained 600 watts. I’ve seen some issues of cables so hot using the wire view pro on shunted 5090’s, that melting is imminent. I would run just a straight cable to your 5090. Return the wire view pro. I’m totally blind on my cable. I had a sensor on it, but I trust it now. It’s been great and that’s inside a closed case, so I let the sensor monitor my case temp instead.

I've have two of them already and the first one was the initially released product from a long time ago and I haven't had any problems with them on both 4090s. The only reason why I had to buy a new one is because my connector is reversed on the Zotac. I think the possibility of the connector melting in the GPU if it is plugged in all the way could be less than a cable melting because all of the current is passing through a large piece of metal instead of individual wires. I think the possibility of the cable melting where it plugs into the WireView is exactly the same as it would be with the cable connected to the video card.

 

The biggest issue with them is not fitting everything that uses this power socket. There are too many design variances from one GPU to the next. I had to use a Dremel tool to carve away part of the acrylic on both of my 4090 Bykski waterbocks. And neither one of them would fit with the stock air cooler installed.

 

From any angle that I look at it there's nothing good or satisfying about this new power connector and it amazes me that they continue using it in spite of its extremely stupid, fragile design and glaring engineering defects. That is a reflection of what a sad, sad, sad situation we have in the PC tech space. We have been overrun by retarded baboons that have been given free reign to make stupid decisions that screw things up for everyone and no accountability for their stupidity.. it is the perfect example of the classic definition of insanity... Continue making the same mistake and expecting a different outcome.