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

[Mr. Fox]

28 minutes ago, jaybee83 said:

wait wat?! ive seen reports of the new type 4090 with max 1.07V but they cant even be crossflashed? seriously? 🫠

boy ive never been so glad to be on the early adopter / bleeding edge... ill cherish my suprim X! 😅

I wonder if SPI flashing would brick the new castrated GPUs? It's so extremely stupid that they would reduce the voltage because of the crappy power connector. The voltage was already too low for good overclocking, so they just screwed it up more by lowering it further. The 4090 needs about 1.300V to do what I want it to do, and there is no reason our Suprim X cards can't handle that with ease. I benched the 3090 KPE with between 1.300 and 1.400V without issues. It didn't need the chintzy 12VHPWR connector for that either. The whole thing is just a fiasco because the Green Goblin felt compelled to do something new and stupid without giving it enough thought and research to determine that it was a stupid idea. Change purely for the sake of change is always idiotic, and this is a very good example of that. The 12VHPWR connect is totally pointless and unnecessary. All it has done is cost money for revised PSU/cables that were not necessary, caused stress, anxiety and frustration, and wasted multiple millions of dollars and hours for GPU owners, GPU manufacturers and GPU repairers.

 

Remember kids: Newer is always better newer.

[jaybee83]

11 minutes ago, Mr. Fox said:

I wonder if SPI flashing would brick the new castrated GPUs? It's so extremely stupid that they would reduce the voltage because of the crappy power connector. The voltage was already too low for good overclocking, so they just screwed it up more by lowering it further. The 4090 needs about 1.300V to do what I want it to do, and there is no reason our Suprim X cards can't handle that with ease. I benched the 3090 KPE with between 1.300 and 1.400V without issues. It didn't need the chintzy 12VHPWR connector for that either. The whole thing is just a fiasco because the Green Goblin felt compelled to do something new and stupid without giving it enough thought and research to determine that it was a stupid idea. Change purely for the sake of change is always idiotic, and this is a very good example of that. The 12VHPWR connect is totally pointless and unnecessary. All it has done is cost money for revised PSU/cables that were not necessary, caused stress, anxiety and frustration, and wasted multiple millions of dollar and hours for GPU owners, GPU manufacturers and GPU repairers.

dont forget the most important factor: it looks fancier with just one cable exiting the gpu 😋 (lets not mention the adapter cable mess that comes after lulz)

 

and true, suprim X is among the cards with the most oversized 4090 vrms, both in number and individually rated amperage, should be able to take quite a beating before it gives out on us...

[Mr. Fox]

6 minutes ago, jaybee83 said:

dont forget the most important factor: it looks fancier with just one cable exiting the gpu 😋 (lets not mention the adapter cable mess that comes after lulz)

 

and true, suprim X is among the cards with the most oversized 4090 vrms, both in number and individually rated amperage, should be able to take quite a beating before it gives out on us...

The only 4090 more overbuilt than the Suprim is the Galax HOF. It's equally castrated as the cheapest models if you can't get your hands on the vBIOS and software that it needs to run wild and free.

[Papusan]

31 minutes ago, Mr. Fox said:

Change purely for the sake of change is always idiotic, and this is a very good example of that. The 12VHPWR connect is totally pointless and unnecessary.

If they absolutely had to change it.... Only Galax/KFA2 made the correct solution for their HOF cards.... Dual 12VHPWR connectors. And mounted them the correct way.

 

 

[Mr. Fox]

6 minutes ago, Papusan said:

If they absolutely had to change it.... Only Galax/KFA2 made the correct solution for their HOF cards.... Dual 12VHPWR connectors. And mounted them the correct way.

 

 

The change was pointless. It was totally unnecessary. My shunt-modded 2080 Ti FTW3 and 3090 KPE both pulled way more power than the 4090 and they were absolutely fine/flawless with three 8-pin connectors. It was a really scummy change to make. I guess they wanted to look smart and be special, but it only made them look extra stupid.

[Papusan]

5 hours ago, Mr. Fox said:

The change was pointless. It was totally unnecessary. My shunt-modded 2080 Ti FTW3 and 3090 KPE both pulled way more power than the 4090 and they were absolutely fine/flawless with three 8-pin connectors. It was a really scummy change to make. I guess they wanted to look smart and be special, but it only made them look stupid.

I think the change was only because Nvidia tried hard to make minimalist PCBs. They are retarded designed. And the lower end SKUs have even smaller PCB. One cute tiny connector fits them all.......

 

 

Then you have the PCI-Sig specs for 8-pin connectors. We could never got more than 525W as maximum with regular firmware (including the 75W max from the Pcie slot). Because we e would never got 4x8-pins connectors from nvidia. Too many have lower end PSUs or PSUs with not proper amount 8 pins connectors. With the Hof cards we are able to get 666W firmware because of 2x12VHPWR. And quite a few can cross-flash this vbios on heavily castrated cards. At least if you have the old 4090 chips from Nvidia. With the new silicon revision for 4090 cards you are screwed. 
 

Just look at AMD. 3x8-pins is all what you’ll get. And we all know that Nvidia and AMD is capable to fully lock any tweaks for TDP and voltage. Wouldn’t that be nice? Locked down to 525w on any cards due PCI-SIG and 3x8pins? I would rather take that 666W firmware and the tiny 12VHPWR connectors Nvidia jumped on any day.

 

Everything going the wrong way with new modern tech. 
 

Nvidia prefers professionals over gamers… Bigger profits. Why waste a full fat AD102 die on 4090Ti cards?

 

https://www.tomshardware.com/news/nvidia-rtx-4500-5000-workstation-cards

[ryan]

cute pic? that looks like me on a good day 😛

 

@tps3443 posted a video a month ago looks like he's running 8800mhz ddr5 which is incredible. hes getting awesome performance out of his system, I wonder how long the 3090 will be relevant for?

 

after doing some quick math on performance of my tweaked system, im getting the equiv of 14000 timespy GS in games. I get half of a 3090s performance, I don't really care why but its nice getting this kind of in game performance. I think my next step will be searching for modified nvidia drivers or just do it myself

 

.all kidding aside I think if I save 500 a month in 6months ill be able to get a discounted 13900k and 4090. and then when the 5k gpus are out ill be able to upgrade to that, will be interesting to see how far I can push desktop parts. The elite and their special samples will always hold the top spots but if I can get into the top 100 i'll be happy.

[Mr. Fox]

 

 

[Papusan]

1 hour ago, Mr. Fox said:

 

 

Nice. I waited a long time to buy a brand new GPU. I skipped buy new graphics cards the last 3.5 years due Mining, scalpers and shortage. Maybe I jumped in at the correct time. In between two sad periods for gamers. And 4090's haven't sinked in price since launch date so not a big loss. And I avoided spend my hard earned money on massive overpriced 4080s and castrated low/mid end Nvidia cards. Yep, maybe the jump on the 4090 HOF was a good purchase/and decision. Hmmm, foods for thoughts bro @Mr. Fox @electrosoft @Rage Set++++

 

Same as with my great mind that said... Avoid custom 12VHPWR adapters/new ATX 3.0 PSU with the fragile tiny 12+4 pin terminal on the PSU panel (what you already have is better). Yep, I prefer trust my own mind/own way thinking tech. 

[Mr. Fox]

2 hours ago, Papusan said:

Nice. I waited a long time to buy a brand new GPU. I skipped buy new graphics cards the last 3.5 years due Mining, scalpers and shortage. Maybe I jumped in at the correct time. In between two sad periods for gamers. And 4090's haven't sinked in price since launch date so not a big loss. And I avoided spend my hard earned money on massive overpriced 4080s and castrated low/mid end Nvidia cards. Yep, maybe the jump on the 4090 HOF was a good purchase/and decision. Hmmm, foods for thoughts bro @Mr. Fox @electrosoft @Rage Set++++

 

Same as with my great mind that said... Avoid custom 12VHPWR adapters/new ATX 3.0 PSU with the fragile tiny 12+4 pin terminal on the PSU panel (what you already have is better). Yep, I prefer trust my own mind/own way thinking tech. 

I think we (several of us) got ours at a good time. Prices were too high, but if they go higher that will prove to be true. It is really sad days for PC enthusiasts. Maybe now is the time we should find something else to do with our time and money.

[Papusan]

40 minutes ago, Mr. Fox said:

I think we (several of us) got ours at a good time. Prices were too high, but if they go higher that will prove to be true. It is really sad days for PC enthusiasts. Maybe now is the time we should find something else to do with our time and money.

Yep, sad times when you have to pay $2,002 USD for a Strix 4080😵 And this isn't even scalper price from the retail shop. Just the normal price you have to pay for graphics cards that should cost the half included the dreaded 25% tax. 

 

 

[Etern4l]

11 hours ago, Mr. Fox said:

 

 

Jay should stick to talking about things he has at least a remote idea about. Unlike with crypto, lower end GPUs are next to useless when it comes to the large  language models (which is what is meant by “AI” in the media right now) - not least because NVidia cut out NVlink support in 4000 desktop series. That stuff needs a lot of VRAM.

Clickbaits pay the bills I guess.

 

The real negative aspect on consumers is caused by the demand on the data centre side. If the demand wasn’t there NVidia would be able to offer better products at better prices to the consumer. There is no such thing as free AI folks.

[jaybee83]

9 hours ago, Mr. Fox said:

 

 

 

8 hours ago, Papusan said:

Nice. I waited a long time to buy a brand new GPU. I skipped buy new graphics cards the last 3.5 years due Mining, scalpers and shortage. Maybe I jumped in at the correct time. In between two sad periods for gamers. And 4090's haven't sinked in price since launch date so not a big loss. And I avoided spend my hard earned money on massive overpriced 4080s and castrated low/mid end Nvidia cards. Yep, maybe the jump on the 4090 HOF was a good purchase/and decision. Hmmm, foods for thoughts bro @Mr. Fox @electrosoft @Rage Set++++

 

Same as with my great mind that said... Avoid custom 12VHPWR adapters/new ATX 3.0 PSU with the fragile tiny 12+4 pin terminal on the PSU panel (what you already have is better). Yep, I prefer trust my own mind/own way thinking tech. 

 

2 hours ago, Etern4l said:

Jay should stick to talking to things he has at least a remote idea about. Unlike with crypto, lower end GPUs are next to useless when it comes to the large  language models (which is what is meant by “AI” in the media right now) - not least because NVidia cut out NVlink support in 4000 desktop series. That stuff needs a lot of VRAM.

Clickbaits pay the bills I guess.

 

The real negative aspect on consumers is caused by the demand on the data centre side. If the demand wasn’t there NVidia would be able to offer better products at better prices to the consumer. There is no such thing as free AI folks.

 

ah good that u spelled it out @Etern4l, i saw the video pop up but hadnt watched it yet. figured it had to do with AI though.

and youre right, AI is just smth for highend gpus with tons of vRAM (unless someone comes up with the idea to be able and distribute loads across lower end gpus with less vRAM then were all done for 😅)

 

also agreed on the indirect impact on lower end gpus, nvidia just doesnt give af about gaming gpus currently, they just throw castrated trash onto the market, slap on a ridiculous price tag and are like: take it or leave it, all the same to us.

im just hoping that consumers will be able to HODL the line and not buy into it. all things considered, i cant imagine shareholders being too kind on literally NOTHING selling on the consumer market, even if itll have a smaller impact on the overall bottom line vs. before.

 

plus, what are they gonna do with all the unsold gpus stored somewhere skyhigh in warehouses? seriously, id bet nvidia would rather just scrap them than selling them at low pricing....ugh bastards

[electrosoft]

Jufes melting down over his 7800X3D problems. I'm not suffering from any of that nonsense but he is on a Crosshair x670e while I'm on MSI.

 

87.1ns latency in the thumbnail is being disingenuous at best. Tuned properly, you should be in the mid to low 60s or high 50s.

 

@jaybee83 have you encountered any of the issues he is flipping out about? I did see in the comments someone recommended disabling TB/USB4.

 

 

[jaybee83]

39 minutes ago, electrosoft said:

Jufes melting down over his 7800X3D problems. I'm not suffering from any of that nonsense but he is on a Crosshair x670e while I'm on MSI.

 

87.1ns latency in the thumbnail is being disingenuous at best. Tuned properly, you should be in the mid to low 60s or high 50s.

 

@jaybee83 have you encountered any of the issues he is flipping out about? I did see in the comments someone recommended disabling TB/USB4.

 

 

 

hm not that im aware. i had my previous 6600 kit tuned in to 6200 and tightened timings, ending up with 55 ns latency. with regards to general latency and bandwidth variations, i have seen somewhat worse results when mclk and fclk are not properly aligned, i.e. im getting the best results at a ratio of 3:1 (at 6200 Mhz that would equate to 2067 Mhz IF). most people ive seen just crank up IF to the max stable (2133 to 2200 Mhz depending on chip quality) since the general AMD stance is that synchronization doesnt matter anymore as with Zen 3 for example. but that aint the case, it does matter much less but theres still an impact.

in any case, might have to do with his setup or smth particular with X3D parts. but seeing as ure not seeing the same issues, probably not. 

 

can u elaborate what exactly hes seeing? dont have wifi here and only limited amount of mobile data 😄 

[Mr. Fox]

Jufes the Dufes, LOL. Always lots of drama, even when the message is right. 

 

On the topic of overpriced GPUs, AI disinformation, and all of the other goof-ball nonsense, it wouldn't surprise me if some of what we are seeing is collusion and some kind of nefarious underlying agenda to "fundamentally transform" (Obama-speak for deliberately screwing up and perma-breaking) gaming so that all gaming is done on a controlled device, with controlled results and price controls. They've already got all of the console jockeys under their thumb. Maybe they think it is time for PC gamers to be subdued into the life of a zombie muppet just like the console sheeple.

 

They can't have us clinging to GPUs, guns or religion. After all, if you have assembled a gaming PC, "you didn't build that." Someone else made it happen.

[electrosoft]

1 hour ago, jaybee83 said:

 

hm not that im aware. i had my previous 6600 kit tuned in to 6200 and tightened timings, ending up with 55 ns latency. with regards to general latency and bandwidth variations, i have seen somewhat worse results when mclk and fclk are not properly aligned, i.e. im getting the best results at a ratio of 3:1 (at 6200 Mhz that would equate to 2067 Mhz IF). most people ive seen just crank up IF to the max stable (2133 to 2200 Mhz depending on chip quality) since the general AMD stance is that synchronization doesnt matter anymore as with Zen 3 for example. but that aint the case, it does matter much less but theres still an impact.

in any case, might have to do with his setup or smth particular with X3D parts. but seeing as ure not seeing the same issues, probably not. 

 

can u elaborate what exactly hes seeing? dont have wifi here and only limited amount of mobile data 😄 

 

Agreed that with everything said and done latency is usually ~54-63ns on average when properly tuned....of course with memory that is qualified to handle those timings, potential Vmem and decent heat dissipation. I am routinely dismayed at XMP/EXPO timings and results. This generation across AMD (AM5) and Intel (12th+) memory OC is the new meta.

 

I think I mentioned it before, but definitely keeping AMD recommended ratios yields the best results or at best outside of their ratios isn't worth the hassle. I ended up scaling back to 3:1 across all testing and usage when everything was said and done.

 

Jufes is ranting about responsiveness, freezing and slow program launching on his Crosshair. I don't use Asus AMD boards nor have I encountered any of his problems on MSI but I know you have one so I was wondering if you had encountered any of his issues. One suggestion was to disable TB/USB4 to fix the problems. Do you have yours disabled or run it enabled?

 

Any performance hits I've ever experienced came down to needing to dial in the memory manually and as always disable  Integrated Radeon Graphics in DM along with install AMDs official chipset drivers and all is well.

[Mr. Fox]

1 hour ago, electrosoft said:

One suggestion was to disable TB/USB4 to fix the problems.

Enabling this crap on my Apex causes memory overclock instability and high latency. In some cases running an AIDA64 memory and cache benchmark will cause a reboot with it enabled. I don't need to worry about disabling it on the Dark mobo because it doesn't have this worthless garbage.

[jaybee83]

1 hour ago, electrosoft said:

Jufes is ranting about responsiveness, freezing and slow program launching on his Crosshair. I don't use Asus AMD boards nor have I encountered any of his problems on MSI but I know you have one so I was wondering if you had encountered any of his issues. One suggestion was to disable TB/USB4 to fix the problems. Do you have yours disabled or run it enabled?

 

Any performance hits I've ever experienced came down to needing to dial in the memory manually and as always disable  Integrated Radeon Graphics in DM along with install AMDs official chipset drivers and all is well.

 

42 minutes ago, Mr. Fox said:

Enabling this crap on my Apex causes memory overclock instability and high latency. In some cases running an AIDA64 memory and cache benchmark will cause a reboot with it enabled.

 

huh, that stuff sounds weird! ive always had TB/USB4 enabled thus far, same for the igpu. the only responsivess issues ive encountered here and there was with excel, but i know for a fact that my current office install is wonky, just havent come around to fixing it yet. and yes, got AMD chipset drivers installed, although i wager thats more important on X3D chips for sheduling.

bro Fox, interesting to see that Intel boards face a similar issue with TB/USB4? 

 

maybe Jufes has armory crate installed, with that cancer on his system ANYthing is possible 😂

[Mr. Fox]

2 minutes ago, jaybee83 said:

maybe Jufes has armory crate installed, with that cancer on his system ANYthing is possible 😂

Amen to that. Since he is a self-declared expert at just about everything, he might have it installed. People that make such claims are usually numbered among the least intelligent or the most lacking in common sense.

[jaybee83]

4 minutes ago, Mr. Fox said:

Amen to that. Since he is a self-declared expert at just about everything, he might have it installed. People that make such claims are usually numbered among the least intelligent or the most lacking in common sense.

i admit with my first win11 install i tried it. i figured hey, gotta experience it for myself before i can make any judgements. boy was i in for a ride, that software is huuuuuuge, so many processes, registry keys and basically impossible to completely remove again... yeah no thx ^^

[Papusan]

1 hour ago, Mr. Fox said:

Enabling this crap on my Apex causes memory overclock instability and high latency. In some cases running an AIDA64 memory and cache benchmark will cause a reboot with it enabled. I don't need to worry about disabling it on the Dark mobo because it doesn't have this worthless garbage.

We can talk about crap/garbage. Enjoy why there is so many repairs on AMD cards.... See especially at 2:25 timestamp.

 

This is why modern AMD cards are so hot....

[Papusan]

19 hours ago, Papusan said:

In between two sad periods for gamers. And 4090's haven't sinked in price since launch date so not a big loss. And I avoided spend my hard earned money on the massive overpriced 4080s and the castrated low/mid end Nvidia cards. Yep, maybe the jump on the 4090 HOF was a good purchase/and decision. Hmmm, foods for thoughts bro @Mr. Fox @electrosoft @Rage Set++++

 

Yep, No one should buy the 4060Ti@8GB - 4060Ti@16GB  or the 4070.

 

 

44:15

 

[Mr. Fox]

49 minutes ago, Papusan said:

We can talk about crap/garbage. Enjoy why there is so many repairs on AMD cards.... See especially at 2:25 timestamp.

 

This is why modern AMD cards are so hot....

Sometimes the simplest solution is to do something else rather than fix someone else's messes. One of the reasons I don't waste my time and money on turdbooks. There is no real fix for poor QC. Lipstick on a pig.

  

36 minutes ago, Papusan said:

Yep, No one should buy the 4060Ti@8GB - 4060Ti@16GB  or the 4070.

While I agree, I'd probably spend $400, $500 or $600 on one of those crappy GeFarts turd GPUs before I would spend a similar amount on a Radeon card. I am not saying that to be offensive to anyone that likes them. I just don't really like the brand. Too many compromises. But, that doesn't mean I think the 4060, 4060 Ti or 4070 GPUs are any good either. Lesser of two evils. At the end of the day, we are in a scenario of being flooded with garbage and having somewhere between zero and one or two good options available to choose from, and the one or two good options are grossly overpriced (a poor value) and still far from resembling an effort to achieve excellence.

• Intel is dishonest
• NVIDIA is wicked
• AMD is stupid
• Micro$lop is all of the above, and ironically; they are the tail that is wagging the dog

The bad options are also overpriced, but within the means (reach) of the victims targeted at their respective price points. There is an option for being screwed at every wallet size, but no great options at any price point.

[Mr. Fox]

Off topic. Mrs. Fox and I have our tenth grandchild (sixth grandson), born this afternoon. I think Brother @Papusanis going to like the Viking names. Magnus Olave Fox. Weighing 7 lbs 4 oz and 20 in tall. I had to Google the names to get the meaning for Magnus and Olave.