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

[electrosoft]

20 minutes ago, Papusan said:

Then they will be sued for false advertising. Both Nvidia and their AIC partners will have even more to think about😄 If they can't fix this the proper way they are in deep trouble. 

 

 

What you say reminds me all too much about Dell's AREA-51M jokebook flagship fiasco from Dell Alienware. Dell was forced to castrate the firmware to stop their trashbooks from burning up. 

 

I am sure they are legally covered somehow if they have to nerf it. They wouldn't be the first or last to cap performance/power for whatever reasons. AMD did it with the 5000 series and initial OC/boost performance with a subsequent release of Agesa.

 

 

[Papusan]

8 minutes ago, johnksss said:

I had to check my second card and my same cable for smoke, fire, brimstone and melting effects on the tips of my 12VHPWR cable and nothing as of yet. Even held my fingers on the cable to check the heat of the wires them selves and the over clock GODs have said.... Nope! Not yet. You are still "SAFE", for now. 😂

Me thinking a lot of the fire is due the new modern connectors not propperly connected. They need to be perfectly connected. JonnyGuru almost killed many of the adapers and proper 12VHPWR connectors/cables in his testing. Almost killed them and still no failures. And he have done testing several months and tested many of them. I imagine this mess is hard for Nvidia to try explain the gamers7consumers. A new mounting manual in the GPU package won't fix this. Many don't seems to understand how it important this is to follow. Gaming first, then fire, LOOL

 

And this is from AMDs HQ looking into reddit threads right now....

[Rage Set]

2 hours ago, cylix said:

 

 

 

HAHAHA, only der8auer throws a fan on the VRAM while drawing over 700W on ambient "water cooling". I guess you can be a little crazy when you get the products for free. 

[Etern4l]

5 minutes ago, Rage Set said:

 

HAHAHA, only der8auer throws a fan on the VRAM while drawing over 700W on ambient "water cooling". I guess you can be a little crazy when you get the products for free. 

 

Surprisingly, he also did a (a bit basic) video on how to optimise the 13900k for energy efficiency. The man has layers  and/or perhaps energy efficiency is a particularly hot topic in Germany if not worldwide at the moment 😉

 

 

 

[tps3443]

11 minutes ago, Etern4l said:

 

Surprisingly, he also did a (a bit basic) video on how to optimise the 13900k for energy efficiency. The man has layers  and/or perhaps energy efficiency is a particularly hot topic in Germany if not worldwide at the moment 😉

 

 

 

It’s fun tuning out CPU’s for efficiency. I always had more fun doing that, then even overclocking them to the moon. Both are fun though.
 

My setup= 230 watts@41K R23

Intel defaults Small Air cooler 252 watts@35K R23

Intel defaults Larger tower air cooler 288 watts@38K R23

intel defaults Customer cooling= unlimited power@41K R23

[johnksss]

22 minutes ago, Rage Set said:

 

HAHAHA, only der8auer throws a fan on the VRAM while drawing over 700W on ambient "water cooling". I guess you can be a little crazy when you get the products for free. 

I sure hope it gets better than that with 700W! I really hope it's because of the 8X lane and not something having to do with how the card was constructed. People are already doing past 3100 with out modding anything right now. And 134 is low for that many watts being pulled.

[Mr. Fox]

5 hours ago, cylix said:

 

 

That would not have required a mod firmware if there were still such a thing as a Kingpin GPU. 1000W power limit and unlocked voltage would have been taken for granted with the standard XOC vBIOS. Sad days with us being left with nothing but gamerboy crap.

[Papusan]

41 minutes ago, johnksss said:

I sure hope it gets better than that with 700W! I really hope it's because of the 8X lane and not something having to do with how the card was constructed. People are already doing past 3100 with out modding anything right now. And 134 is low for that many watts being pulled.

Have seen several with exactly same problems. Could be sBios/Mb don’t play well with the cards. People start up and everything works well with the card running with all 16 lanes then come back and see only 8. Very random. And sometimes it is stuck with 8 lines all times. Mostly seen on Z690 boards. 

[Etern4l]

2 minutes ago, Papusan said:

Have seen several with exactly same problems. Could be sBios/Mb don’t play well with the cards. People start up and everything works well with the card running with all 16 lanes then come back and see only 8. Very random. And sometimes it is stuck with 8 lines all times. Mostly seen on Z690 boards. 

 

In some rare cases people may have missed the small print in their mobo documentation. For example, the MEG ACE downgrades the first PCie slot to 8 lanes if a particular M.2 slot is in use.

[johnksss]

11 minutes ago, Papusan said:

Have seen several with exactly same problems. Could be sBios/Mb don’t play well with the cards. People start up and everything works well with the card running with all 16 lanes then come back and see only 8. Very random. And sometimes it is stuck with 8 lines all times. Mostly seen on Z690 boards. 

Yes, I have ran across this issue myself, but it hasn't happened in a while. Talking about being stuck on 8 and 4 lanes. I had to pull the card reset the bios and boot in a different PCIE, then put it back and eventually 16 lanes came back.

 

And I think you might be right with something going on with the bios....I'm trying to work through it now to see if it is somewhat like how it was with the 20 series in the laptops.

[Reciever]

24 minutes ago, Mr. Fox said:

That would not have required a mod firmware if there were still such a thing as a Kingpin GPU. 1, 000W and unlocked voltage would have been taken for granted with the standard XOC vBIOS. Sad days with us being left with nothing but game boy crap.

All roads lead back to the soldering iron. 

 

Recently had to fix my own tools at work since the "engineering department" only manages to scratch off the pad rendering it destroyed. Since I had been following a fair number of exploits by various NBR members I knew I could lightly scratch the covering for the trace and soldered a wire to the pin for my usb port. Fully operational. 

 

Sometimes I question engineers lack of motivation. Career specialist best by a 10 dollar iron and 10 minutes of youtube lol

 

Though I did try to use solder wick as solder and without paste, while not plugged in. Baby steps.

 

Next little project is soldering a 4000Mah battery to my homebrew PSP, I think its supposed to be good for 15+ hours of continuous use after that. Not my idea of course.

[Talon]

 

AMD confirms 7900 XTX is 4080 competitor, not 4090. Damn crazy to hear AMD come right out and say they have no answer for the 4090. 
 

If it’s competitive the 4080 that will be good for all, and I see Nvidia dropping prices a tad in the near future, at least on the mid range cards since AMD has bowed out of the high end now. 

[ryan]

nvidia in 10 years..amish mafia at its best

 

 

just listen to the lyrics and see the simile or metaphor

[Papusan]

45 minutes ago, Talon said:

 

AMD confirms 7900 XTX is 4080 competitor, not 4090. Damn crazy to hear AMD come right out and say they have no answer for the 4090. 
 

If it’s competitive the 4080 that will be good for all, and I see Nvidia dropping prices a tad in the near future, at least on the mid range cards since AMD has bowed out of the high end now. 

LOOL. AMD wanted create a card that fits well in a small Dell Alienware chassis. HaHaaaaaaa. Maybe the idea come from Azor? The future is SFF. But Nvidia have no cards to compete with at this price point. Will be fun to see if nvidia change 4080 prices before or after the launch. Maybe they will wait after the launch of the AMD cards to see where the 4080 should be priced? 

 

Edit. Nvidia have now the golden chance to go back to Titan branding to max out profits from the full fat ADA and just let the down cut 4090 be the fastest geforce gaming card at +1600$. 

[johnksss]

2 minutes ago, Papusan said:

LOOL. AMD wanted create a card that fits well in a small Dell Alienware chassis. HaHaaaaaaa. Maybe the idea come from Azor? The future is SFF. But Nvidia have no cards to compete with at this price point. Will be fun to see if nvidia change 4080 prices before or after the launch 4080. Maybe they will wait after the launch of the AMD cards to see where the 4080 should be priced?

You were not lying when you said AMD was laughing!! 😂😂😂

[electrosoft]

5 hours ago, Talon said:

 

AMD confirms 7900 XTX is 4080 competitor, not 4090. Damn crazy to hear AMD come right out and say they have no answer for the 4090. 
 

If it’s competitive the 4080 that will be good for all, and I see Nvidia dropping prices a tad in the near future, at least on the mid range cards since AMD has bowed out of the high end now. 

 

Just because he says it targets the 4080 (easy win?) doesn't mean it also won't compete with the 4090. But with AMD numbers to go by only, we will see. ADA didn't match the hype either with people saying it was going to be 95% faster in games and on average it was ~60% overall. I expect a similar reality check when the real reviews come in for the 7900xtx. That CP2077 1.7x uplift is potentially going to be a major outlier.

 

Nvidia will retain the performance crown overall (when all was said and done just as the 3090ti did for Ampere against the 6950xt)  but they will have to re-evaluate their entire stack now with AMD bringing so much performance in the <=$1000 segment.

 

If you can potentially get 90% of the performance of a 4090 for $600 less in pure Raster?

 

RT is going to be better (ish) than Ampere but 4090 just decimates it....

DLSS/FSE/XESS? Meh.....

 

Still irked over all the leaks that kept pushing the massive SPs but didn't note the 2x per change meaning CU went from 80 to 96 not 160+ and SP is realistically 6144 x 2 not 12288. Threw my guestimations WAY off giving me a big fat WHIFF on this one like a boss. 🙂

 

We'll wait for the real numbers, but we know AMD can and has cherry picked with the best of them but again this will really shake up Nvidia's approach to pricing especially in the lower high / upper mid and down stack.

 

Or

 

OR

 

 

We're going to end up with another 5700xt which was a great card but simply couldn't compete with the 2080ti overall but put plenty of heat on Nvidia in the upper middle on down.

 

4 hours ago, Papusan said:

LOOL. AMD wanted create a card that fits well in a small Dell Alienware chassis. HaHaaaaaaa. Maybe the idea come from Azor? The future is SFF. But Nvidia have no cards to compete with at this price point. Will be fun to see if nvidia change 4080 prices before or after the launch. Maybe they will wait after the launch of the AMD cards to see where the 4080 should be priced? 

 

That would be the smart play from Nvidia for 4080 pricing.

 

 

[Papusan]

15 minutes ago, electrosoft said:

That would be the smart play from Nvidia for 4080 pricing.

If you can potentially get 90% of the performance of a 4090 for $600 less in pure Raster?

My edit above. Nvidia have now the golden chance to go back to Titan branding to max out profits from the full fat ADA and just let the down cut 4090 be the fastest geforce gaming card at +1600$. 

 

And nvidia only need to change 4080 price point now to fit in the 4080 Ti. 

 

What aboth 4080Ti@1299$ and $1099 for 4080 if the performance difference between 4080 and AMD cards isn't huge. Nvidia can upsell their better features if 4080 lack the full power to beat AMD overall. People may agree to pay 99$ for Nvidia's better features as long the card perform around the same.

[Reciever]

19 minutes ago, electrosoft said:

 

Just because he says it targets the 4080 (easy win?) doesn't mean it also won't compete with the 4090. But with AMD numbers to go by only, we will see. ADA didn't match the hype either with people saying it was going to be 95% faster in games and on average it was ~60% overall. I expect a similar reality check when the real reviews come in for the 7900xtx. That CP2077 1.7x uplift is potentially going to be a major outlier.

 

Nvidia will retain the performance crown overall (when all was said and done just as the 3090ti did for Ampere against the 6950xt)  but they will have to re-evaluate their entire stack now with AMD bringing so much performance in the <=$1000 segment.

 

If you can potentially get 90% of the performance of a 4090 for $600 less in pure Raster?

 

RT is going to be better (ish) than Ampere but 4090 just decimates it....

DLSS/FSE/XESS? Meh.....

 

Still irked over all the leaks that kept pushing the massive SPs but didn't note the 2x per change meaning CU went from 80 to 96 not 160 and SP is realistically 6144 x 2 not 12288. Thew my guestimations WAY off giving be a big fat WHIFF on this one like a boss. 🙂

 

We'll wait for the real numbers, but we know AMD can and has cherry picked with the best of them but again this will really shake up Nvidia's approach to pricing especially in the lower high / upper mid and down stack.

 

Or

 

OR

 

 

We're going to end up with another 5700xt which was a great card but simply couldn't compete with the 2080ti overall but put plenty of heat on Nvidia in the upper middle on down.

 

 

That would be the smart play from Nvidia for 4080 pricing.

 

 

It could also be that AMD is trying to deflate the hype train. Honestly considering how many times people expect AMD to take them to the moon only to be shown the observatory has hurt their image just as much as their marketing for Vega did.

[Mr. Fox]

If it comes out of Azor's mouth then it should be considered a lie until proven otherwise. AMD hiring a worthless snake like that clown makes them a less trustworthy company and harms their brand credibility on face value.

  

36 minutes ago, Papusan said:

The future is SFF.

 

Not mine... I've got no use for compromised garbage like that. Gateway to turdbook city in a non-portable form factor. No thank you... hard pass.

[Ashtrix]

Well I lost respect to this Steve.

 

 

AMD Radeon coverage is not well rounded and cringe. Goes off on how AMD misleading with fake 8K, yes it is FSR obv it's not real it is fake indeed and they got access to slide deck because these Techtubers go on a trip sponsored by all corporations, which is what TechPowerUp and Anandtech post. Anyways so I went and checked his Nvidia coverage, well not a single word on DLSS green bars that Nvidia showed. Here's his Nvidia coverage for Ada RTX 4000 cards.

 

Then how AMD did a jab at Nvidia on power adapter.. and it's not good in his own words lmao. Well AMD is using the free opportunity and the 12VHPWR is a joke no matter what. Saying, how we can also use native gen 5 power adapter (yea just ignore the adapter in the box of a $1600 card and buy others, what about non-native ATX3.0 PSUs ? ah lets forget it right). Boy how well it aged like milk in a day, when we got native 12VHPWR connectors melting out.

 

Also DP2.0 marketing is also not good as to him and somehow misleading by AMD as we do not see any frames... Hmm a $1600 (4090) card with no DP2.0 heck even Intel ARC has it. AMD X670E boards support it too so do their Ryzen processors. Going by that logic, why have PCIe 5.0 X16 lanes on the dGPU PCIe slot on latest mobos ? we dont have GPUs nothing utilizing them. Since when having more is bad..

 

He should have used this slide instead of wasting time on about fake FSR trash 8K marketing fluff nobody cares about 8K unless they are really pushing it native or using it to bench an SLI rig. Anyways here's the slide LTT mentions it as well btw. This is the exact slide which useful. Not that FSR benchmarks garbage that AMD showed. Same with Nvidia they had only 1 slide which was good it showed that 4080 12GB was losing out to 3090Ti.

 

 

Nvidia slide btw which was worth..notice the 4080 16GB perf vs 3090Ti there.

 

 

A small note - This is me who owns an Nvidia GPU and always owned them only. However that is not going to be in the future though.

 

As for Performance estimates here's a LTT video on the charts (ignore that clickbaity thumbnail lol)

 

Skip video and here's the estimates. Note these are Raster ONLY. He shows the RT ones where AMD falls flat totally as expected. Again except Metro I do not see any worthy title out there. Portal RTX, Quake RT are maybe a couple if you really want since both are fully pathtraced like Metro RT only edition.

 

 

 

That 7900XTX will kick the hell out Nvidia flagship part for 60% less price. RTX 3090 TI vs 4080 16GB there won't be massive gains it will be limited as per Nvidia's own slides only reason left is to spend more cash unless they care about RT and DLSS3. This card will KO the 4080 16GB. And fight against the 4090. AMD will release a 7950XT to fight it again once 4090Ti gets released.

 

Finally that awful Frame Interpolation type thing. AMD Is saying they are working to make their FSR3 across other cards than RDNA3 and say it explicitly by that idiot Frank Azor, note AMD is using AI accelerators in RDNA3 and not RDNA2 so if they use it RDNA2 won't but they are hinting it will get.  Also no confirmation if its Frame Interpolation or not.. hoping it is not.

 

Quote

 

[AMD FSR3] is not a reaction or a quick thing [to DLSS3], it is absolutely something we have been working on for a while. Why is it taking a little bit longer for it come out, that you’d probably hoped for? The key thing to remember about FSR is the FSR philosophy and FSR until now did not just work on RDNA2 or RDNA1 they work on other generations of AMD graphics cards. They also work on competitors graphics cards. It is exponentioally harder than if we just made it work on RDNA3. […] We really do want to work on more than just RDNA3. 

— Frank Azor to PCWorld

 

 

Imagine that fake FPS booster doesn't use anything AI but also works on Ampere, massive kick in the nut for Nvidia for gating that tech to Ada Lovelace.

[tps3443]

Hey, @johnksss how much voltage do you send to these on your MSI Unify-X to use your DDR5 7600 XMP profile daily? Thanks.

CPU VDDQ, and CPU VDD2


download instagram high resolution photos

[ryan]

I was wrong saw a video on ddr5...the speed can make a difference in gaming when your cpu bottlenecked....interesting i did not know that

[electrosoft]

4 hours ago, Reciever said:

It could also be that AMD is trying to deflate the hype train. Honestly considering how many times people expect AMD to take them to the moon only to be shown the observatory has hurt their image just as much as their marketing for Vega did.

 

That WOULD be a refreshing change of pace. Quietly undersell / deflate the hype train and when launched be better than expected.

 

I am sure the lackluster response to AM5 may have played a small part in their approach/pricing.

 

4 hours ago, Mr. Fox said:

If it comes out of Azor's mouth then it should be considered a lie until proven otherwise. AMD hiring a worthless snake like that clown makes them a less trustworthy company and harms their brand credibility on face value.

  

 

Not mine... I've got no use for compromised garbage like that. Gateway to turdbook city in a non-portable form factor. No thank you... hard pass.

 

Ugh, Azor.....destroyer of Alienware....

 

SFF has its place in hierarchy. I like SFF builds but not for my main rig ever.

 

4 hours ago, Ashtrix said:

Well I lost respect to this Steve.

 

 

AMD Radeon coverage is not well rounded and cringe. Goes off on how AMD misleading with fake 8K, yes it is FSR obv it's not real it is fake indeed and they got access to slide deck because these Techtubers go on a trip sponsored by all corporations, which is what TechPowerUp and Anandtech post. Anyways so I went and checked his Nvidia coverage, well not a single word on DLSS green bars that Nvidia showed. Here's his Nvidia coverage for Ada RTX 4000 cards.

 

Then how AMD did a jab at Nvidia on power adapter.. and it's not good in his own words lmao. Well AMD is using the free opportunity and the 12VHPWR is a joke no matter what. Saying, how we can also use native gen 5 power adapter (yea just ignore the adapter in the box of a $1600 card and buy others, what about non-native ATX3.0 PSUs ? ah lets forget it right). Boy how well it aged like milk in a day, when we got native 12VHPWR connectors melting out.

 

Also DP2.0 marketing is also not good as to him and somehow misleading by AMD as we do not see any frames... Hmm a $1600 (4090) card with no DP2.0 heck even Intel ARC has it. AMD X670E boards support it too so do their Ryzen processors. Going by that logic, why have PCIe 5.0 X16 lanes on the dGPU PCIe slot on latest mobos ? we dont have GPUs nothing utilizing them. Since when having more is bad..

 

He should have used this slide instead of wasting time on about fake FSR trash 8K marketing fluff nobody cares about 8K unless they are really pushing it native or using it to bench an SLI rig. Anyways here's the slide LTT mentions it as well btw. This is the exact slide which useful. Not that FSR benchmarks garbage that AMD showed. Same with Nvidia they had only 1 slide which was good it showed that 4080 12GB was losing out to 3090Ti.

 

 

Nvidia slide btw which was worth..notice the 4080 16GB perf vs 3090Ti there.

 

 

A small note - This is me who owns an Nvidia GPU and always owned them only. However that is not going to be in the future though.

 

As for Performance estimates here's a LTT video on the charts (ignore that clickbaity thumbnail lol)

 

Skip video and here's the estimates. Note these are Raster ONLY. He shows the RT ones where AMD falls flat totally as expected. Again except Metro I do not see any worthy title out there. Portal RTX, Quake RT are maybe a couple if you really want since both are fully pathtraced like Metro RT only edition.

 

 

 

That 7900XTX will kick the hell out Nvidia flagship part for 60% less price. RTX 3090 TI vs 4080 16GB there won't be massive gains it will be limited as per Nvidia's own slides only reason left is to spend more cash unless they care about RT and DLSS3. This card will KO the 4080 16GB. And fight against the 4090. AMD will release a 7950XT to fight it again once 4090Ti gets released.

 

Finally that awful Frame Interpolation type thing. AMD Is saying they are working to make their FSR3 across other cards than RDNA3 and say it explicitly by that idiot Frank Azor, note AMD is using AI accelerators in RDNA3 and not RDNA2 so if they use it RDNA2 won't but they are hinting it will get.  Also no confirmation if its Frame Interpolation or not.. hoping it is not.

 

 

Imagine that fake FPS booster doesn't use anything AI but also works on Ampere, massive kick in the nut for Nvidia for gating that tech to Ada Lovelace.

 

"This card will KO the 4080 16GB. And fight against the 4090."

 

This in spades. Based on presentation and performance estimates (after adjustments), the 7900XTX is going to be an excellent card and price:performance monster.

 

 

Linus going into more about AMD's new cards and Nvidia:

 

 

[johnksss]

10 hours ago, tps3443 said:

Hey, @johnksss how much voltage do you send to these on your MSI Unify-X to use your DDR5 7600 XMP profile daily? Thanks.

CPU VDDQ, and CPU VDD2


download instagram high resolution photos

Right now it's 1.4 across the board.

 

Edit: I'm also not using my Unify-X as it's at MSI. I'm using my Hero Z790 Board at the moment.  And it seems to be potentially stable at 7800 for the moment.

[Mr. Fox]

FYI - in case it helps someone else avoid the frustration... user error on my part... sort of. To avoid issues, recycle the same password for both accounts. Being smart and using a distinct password for each account doesn't work. Chrome thinks they are the same.