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

[electrosoft]

KPE 3090ti available direct now:

 

https://www.evga.com/products/product.aspx?pn=24G-P5-4998-KT

 

No surprise there.
 
Every miner that had potentially queued up dropped out
Those wanting a block dropped out
Those who thought the wait was too long dropped out
Those who didn't want to pony up $2500 w/ a PSU dropped out
Those who decided to wait for 4000 dropped out
 
EVGA will keep it like this for awhile and then the next step will be to offer it without the PSU for ~$2299.99 is my next guess. Either that or the known limited run will finish selling out from their store front and that will be that.

[Mr. Fox]

4 hours ago, electrosoft said:

True, if you are only wanting Nvidia regardless of what AMD does then it doesn't matter. Any bit of potentially wanting AMD GPUs was probably thwarted by your 5950x...um....let's call it "adventure." 😅

That, plus their crappy GPU drivers and related software, no DLSS and lousy ray tracing.

4 hours ago, electrosoft said:

KPE 3090ti available direct now:

 

https://www.evga.com/products/product.aspx?pn=24G-P5-4998-KT

 

No surprise there.
 
Every miner that had potentially queued up dropped out
Those wanting a block dropped out
Those who thought the wait was too long dropped out
Those who didn't want to pony up $2500 w/ a PSU dropped out
Those who decided to wait for 4000 dropped out
 
EVGA will keep it like this for awhile and then the next step will be to offer it without the PSU for ~$2299.99 is my next guess. Either that or the known limited run will finish selling out from their store front and that will be that.

I am actually surprised that they released it. I think it was a very bad business decision. I think they should have known/understood that is was going to be an unattractive purchase for many, if not most, of the people that are willing to spend a massive wad of cash to own the best overclocking enthusiast GPU. The lack of a waterblock makes it totally out of the question and a product of zero interest to many of us. And, it would be an almost insane waste of money for anyone that already ponied up the cash for a 3090 KPE. NVIDIA releasing the 3090 Ti was a kick in the nuts to all 3090 owners and EVGA releasing the 3090 Ti KPE (and FTW3) was like a slap in the face to customers, and kicking themselves in the nuts.

[electrosoft]

5 hours ago, Mr. Fox said:

That, plus their crappy GPU drivers and related software, no DLSS and lousy ray tracing.

I am actually surprised that they released it. I think it was a very bad business decision. I think they should have known/understood that is was going to be an unattractive purchase for many, if not most, of the people that are willing to spend a massive wad of cash to own the best overclocking enthusiast GPU. The lack of a waterblock makes it totally out of the question and a product of zero interest to many of us. And, it would be an almost insane waste of money for anyone that already ponied up the cash for a 3090 KPE. NVIDIA releasing the 3090 Ti was a kick in the nuts to all 3090 owners and EVGA releasing the 3090 Ti KPE (and FTW3) was like a slap in the face to customers, and kicking themselves in the nuts.

 

The lack of a block while the FTW3 is getting one really pushed a large contingent of KPE buyers away. The writing was on the wall when they pulled the 3090ti FTW3 HC. It definitely was a kick in the teeth to KPE 3090 owners. I know I assumed (silly me) that when EVGA launched the KPE 3090 that the 3090 was the top dog and that was that. A lot of that is on Nvidia too. Playing games when they should have just released the top tier from the get go. One thing we can say about AMD is they brought everything they had to the table and didn't hold back.

 

I ordered one because I wanted it pure and simple. I like the look of it and I plan on testing their new fans (along with the asetek rev 6 pump vs 5) against my old KPE results and then slap on some AC P12s and see what they bring to the table. I sold my KPE 3090 back in October and the sell price covers this and a little bit on top. I think my KPE 3090 went for $2900.00. With the new EVGA associate code rules and end of the associates program you can just use your own code now and knock 3% off so I saved $75 (!) and no way I was paying for next day or 2 day shipping which is $120 extra when they are offering free ground shipping to elite members. When I bought my KPE 3090 even ground shipping was 50-60. Throw in a spot of cash back with my credit card and here we are.

 

Everyone, feel free to continue to dump on Nvidia and EVGA. 🙂 I made a fully educated purchase of my own free will fully cognizant of market conditions and timing. You won't hurt my feelings and I'll prolly end up chiming in and agreeing with you all....while I install and play around with the KPE 3090ti when it arrives sometime next week.🤑

[Mr. Fox]

25 minutes ago, electrosoft said:

Everyone, feel free to continue to dump on Nvidia and EVGA. 🙂 I made a fully educated purchase of my own free will fully cognizant of market conditions and timing. You won't hurt my feelings and I'll prolly end up chiming in and agreeing with you all....while I install and play around with the KPE 3090ti when it arrives sometime next week.🤑

I can relate to that and I've done that more than a few times myself. When you want something and it's within your means to acquire it sometimes it just doesn't matter whether the price is reasonable or not. That has nothing to do with whether or not it is reasonably priced. Most important is what you expect to extract from it. When the price is high, the bar on minimum expectations goes up with it, as it should. Congrats on the beast GPU, bro.

[ryan]

almost broke 50k in night raid...everytime I bump cpu/gpu clocks score plummets bizzaro

 

[Papusan]

Today’s blooming Joke. Tomorrow’s Intel processors may get 3 different cores. One of those 3 is the baby’s babysitter core, LOOL @Ashtrix

 

Imagine 8+16+16 cores. That will be 40 cores and 48 threads. But a tragedy in thread count due all the E-cores. A fantastic leap in core counts. S why not just scrap the few P-cores and instead fill the silicon up with only small E-cores? If they started use LP E-cores they could probably cram in a lot more cores. Wouldn’t it be nice have a real Intel Threadripper variant as mainstream desktop processors? They could even offer it for laptops.  
 

The tech adoption from Smartphones into desktops/laptops has to stop. And AMD is tempted to do the same as well. Unisex can be nice but not this messed up tech for computers.

https://www.sweclockers.com/nyhet/34572-intel-meteor-lake-till-barbara-far-tre-olika-karnor

 

”It is also the first time that Intel divides the energy-efficient cores into two different variants, where one is referred to as "LP E-Core". It is possible that LP stands for low power and that Intel takes another leaf from ARM and the mobile manufacturers' book for a hybrid configuration with three different sized cores. However, the diagrams only show information about the portable series. How the configuration can be for the stationary variants remains to be seen”

[jaybee83]

18 hours ago, electrosoft said:

 

 

Based on all the leakers and Moore's Law is Dead's last video, it is valid and Nvidia is in total agreement with AIBs to push back 4000 series while the 3000 series are clogging the supply lines and so many used models are saturating the pre-owned market.

 

Interesting tidbit. I was thinking of all the Desktop GPUs I've owned ever since my very first EGA card back in the early 90s I bought from Radio Shack and I've only once bought a used GPU in my entire life and that was a KPE 1080ti because EVGA sold out of them. Not that I'm adverse to buying used it just seemed to work out that way....for over 30yrs.

 

 

currently watching the video, soooo much SCHADENFREUDE, its not even funny anymore 😛 let the vendors, shops and Nvidia buuuuuuuurn after making so much shameful profit these past 2 years, yehaw!

[Ashtrix]

5 hours ago, Papusan said:

Today’s blooming Joke. Tomorrow’s Intel processors may get 3 different cores. One of those 3 is the baby’s babysitter core, LOOL @Ashtrix

 

Imagine 8+16+16 cores. That will be 40 cores and 48 threads. A fantastic leap in core counts. S why not just scrap the few P-cores and instead fill the silicon up with only small E-cores? If they started use LP E- cores they could probably cram in a lot more cores. Wouldn’t it be nice have a real Intel Threadripper variant as mainstream desktop processors? They could even offer it for laptops.  
 

The tech adoption from Smartphones into desktops/laptops has to stop. And AMD is tempted to do the same as well. Unisex can be nice but not this messed up tech for computers.

https://www.sweclockers.com/nyhet/34572-intel-meteor-lake-till-barbara-far-tre-olika-karnor

 

”It is also the first time that Intel divides the energy-efficient cores into two different variants, where one is referred to as "LP E-Core". It is possible that LP stands for low power and that Intel takes another leaf from ARM and the mobile manufacturers' book for a hybrid configuration with three different sized cores. However, the diagrams only show information about the portable series. How the configuration can be for the stationary variants remains to be seen”

 

What a shame really.

 

Intel is really hardcore into the BGA DUMPSTER. Pathetic display of ceding the CPU design prowess and efficiency. They found out this way to rip off ARM design, not only they ruined Desktops but now doubling down on this cancer. Ugh. Now I do not think Intel will stop adding E nonsense even if they release CORE series successor at this point, doesn't matter if 2026 or not. Intel is clearly into this.

 

Also notice bro papu how Windows 11 abandoned Desktop centric UI to a mobile junk. And worse is Pluton processors that block Installation of other OSes thanks to Microtrash's silicon level backdoor. Fresh out of oven news.

 

Microsoft’s Pluton prevents Lenovo ThinkPad laptops with AMD Ryzen 6000 PRO from booting with Linux

Quote

This laptop is equipped with Lenovo exclusive AMD Ryzen PRO 6860Z processor with built-in Microsoft Pluton security co-processors. This is a dedicated chip that is supposed to increase security for Windows systems by verifying UEFI certificate keys. The problem is that it only trusts Microsoft’s key, not any 3rd party UEFI keys that are used by various Linux distributions.

 

Quote

 

“This means that given the default firmware configuration, nothing other than Windows will boot. It also means that you won’t be able to boot from any third-party external peripherals that are plugged in via Thunderbolt. There’s no security benefit to this. If you want security here you’re paying attention to the values measured into the TPM, and thanks to Microsoft’s own specification for measurements made into PCR 7, switching from booting Windows to booting something signed with the 3rd party signing key will change the measurements and invalidate any sealed secrets. It’s trivial to detect this. Distrusting the 3rd party CA by default doesn’t improve security, it just makes it harder for users to boot alternative operating systems.”

— Matthew Garrett, Security developer at Aurora

 

What a disaster ? do people care ? Nope they will simply chug it all down. Windows 11 adoption rate is really faster than 10 imagine even after GWX nagware Windows 10 failed because people thought it was garbage vs 7. Now 11 is used by so many fools.

 

Also there's some good news, AMD's Threadripper is going to wreck havoc on Intel Sapphire Rapids XEON (MIA delayed atm). Genoa is 96C192T beast.

 

AMD EPYC 9000 “Genoa” processor lineup leaks out, up to 96 Zen4 cores and 360W TDP

Quote

 

The full configuration of EPYC Genoa silicon offers up to 96 Cores and 192 threads. Thus far only one production ready processor with such a core count is known, the EPYC 9654P. This CPU would operate in 2.0 to 2.15 GHz frequency and feature 360W TDP by default. However, according to YuuKi’s chart, there is also an engineering sample operating with up to 400W, it may simply be a pre-production sample of the same processor though.

 

The EPYC 9000 series would feature various core configurations, including 96, 64, 48, 32, 24 and even 16 cores. Each of those variants will also have different total cache, ranging from 384MB to just 64MB. The 16-core EPYC 9174F would be the highest clock SKU up to 3.8 GHz but also requiring 320W TDP.

 

AMD Genoa CPUs are set to launch later this year. Officially Zen4 data-center CPUs should become available in the fourth quarter

 

 

 

Zen CCD is evolving while Intel is devolving into this madness of E cores, granted Sapphire Rapids doesn't use E cores and uses EMIB to link 14C24T Tiles x4 still AMD is really pushing x86 to new heights.

 

Still shame since AMD abandoned the HEDT, Intel also. Now with SPR delay and without pressure from Intel AMD won't even bother and on Desktop side many do not even care about HEDT they just want those pathetic trash woke gaming performance not the PCIe or I/O or etc.

[electrosoft]

5 hours ago, jaybee83 said:

currently watching the video, soooo much SCHADENFREUDE, its not even funny anymore 😛 let the vendors, shops and Nvidia buuuuuuuurn after making so much shameful profit these past 2 years, yehaw!

 

What Nvidia (and AMD to an extent) found out was that there are more than enough customers willing to pay but I truly do think it is going to backfire this next cycle as many who upgraded to 3000 are going to skip 4000. Of course Pascal and Turning owners who skipped 3000 will probably upgrade to 4000 and hardcore 3000 owners will upgrade to 4000.

 

The last few years were just a perfect storm of so many things from the Pandemic supply constraints for many reasons to (another) crypto explosion, working from home, societal restrictions = added disposable income and a genuinely worthy upgrade GPU to name a few.

 

Now we are facing a world wide recession, world wide inflation, no crypto-boom (yet) and a heavily saturated Ampere and RDNA2 market. There's a reason Nvidia wanted to cut back on their orders but I also think Nvidia will still increase prices vs last cycle (as they have done the last three cycles) and I expect the 4090 to clock in around $1699.99. We talk about massively over MSRP prices but the fact still remains before any of this happened the MSRP of the 3090 for the Founders addition was $1500 and all other AIB models were equal or greater than that. No way Nvidia is regressing even if it means selling lesser units.

 

But the days of 100% selling out 24/7 are over.

 

 

[Papusan]

43 minutes ago, electrosoft said:

 

What Nvidia (and AMD to an extent) found out was that there are more than enough customers willing to pay but I truly do think it is going to backfire this next cycle as many who upgraded to 3000 are going to skip 4000. Of course Pascal and Turning owners who skipped 3000 will probably upgrade to 4000 and hardcore 3000 owners will upgrade to 4000.

 

The last few years were just a perfect storm of so many things from the Pandemic supply constraints for many reasons to (another) crypto explosion, working from home, societal restrictions = added disposable income and a genuinely worthy upgrade GPU to name a few.

 

Now we are facing a world wide recession, world wide inflation, no crypto-boom (yet) and a heavily saturated Ampere and RDNA2 market. There's a reason Nvidia wanted to cut back on their orders but I also think Nvidia will still increase prices vs last cycle (as they have done the last three cycles) and I expect the 4090 to clock in around $1699.99. We talk about massively over MSRP prices but the fact still remains before any of this happened the MSRP of the 3090 for the Founders addition was $1500 and all other AIB models were equal or greater than that. No way Nvidia is regressing even if it means selling lesser units.

 

But the days of 100% selling out 24/7 are over.

 

 

Nvidia was smart enough to release 3080@10GB as their 4K gaming cards for an ok price (msrp). A nice way try lure people into next gen cards once similar cards (for 4K) is out from 40xx series. Exactly as they do for laptops (castrate the cards to push you over on next gen). Not so sure the next new tricks will help when miners is gone and people have less money in their pockets. Hope many buy used cards as 3080ti/4090 to stop price increase for Ada. All they can do then is offer 4090 cards on this side of Christmas. And they just can’t offer middle range cards if the web is flooded with used cards. Maybe a nice punch from AMD this fall also will help as well on MSRP price increase. If the gamers playing their game correct then we will see Nvidia will begging you to buy their new cards. 
 

As for me, I have skipped last gens cards (only used old cards so no profits from my pocket). Hope I will be in minority buying new cards this fall/spring 2023 🙂 Let the bloodbath start. Rather before than later, LOOL

 

Wouldn't it be nice with a 40% drop in new sales of cards? I’m sure this will help staggering greed. 

[jaybee83]

9 hours ago, Papusan said:

Today’s blooming Joke. Tomorrow’s Intel processors may get 3 different cores. One of those 3 is the baby’s babysitter core, LOOL @Ashtrix

 

Imagine 8+16+16 cores. That will be 40 cores and 48 threads. But a tragedy in thread count due all the E-cores. A fantastic leap in core counts. S why not just scrap the few P-cores and instead fill the silicon up with only small E-cores? If they started use LP E-cores they could probably cram in a lot more cores. Wouldn’t it be nice have a real Intel Threadripper variant as mainstream desktop processors? They could even offer it for laptops.  
 

The tech adoption from Smartphones into desktops/laptops has to stop. And AMD is tempted to do the same as well. Unisex can be nice but not this messed up tech for computers.

https://www.sweclockers.com/nyhet/34572-intel-meteor-lake-till-barbara-far-tre-olika-karnor

 

”It is also the first time that Intel divides the energy-efficient cores into two different variants, where one is referred to as "LP E-Core". It is possible that LP stands for low power and that Intel takes another leaf from ARM and the mobile manufacturers' book for a hybrid configuration with three different sized cores. However, the diagrams only show information about the portable series. How the configuration can be for the stationary variants remains to be seen”

hey papu and fox, how about ONE P core clocked to 6 Ghz paired with 63 flimsy wuss cores = 64 cores in mainstream desktop FTW! (with 65 threads LOL)

just enough to get that single thread crown and "supposedly" win in games 😂

[electrosoft]

12 minutes ago, jaybee83 said:

hey papu and fox, how about ONE P core clocked to 6 Ghz paired with 63 flimsy wuss cores = 64 cores in mainstream desktop FTW! (with 65 threads LOL)

just enough to get that single thread crown and "supposedly" win in games 😂

 

Ugh...........

 

 

[Mr. Fox]

The problem is the perceived importance of being energy efficient and consuming less electrical power to avert an imaginary cataclysmic event and the notion that disaster will follow if they do not. That irrational hoax guides more cockamamie ideas than we can imagine. It is remarkable how often claims of "science" are used to justify an agenda and even more incredible how so many sheeple blindly embrace heresy because somebody that seems smart told them it was based on science.

[Reciever]

I like Energy Efficiency but mostly because I like keeping my money.

 

Though that just means said product is more expensive more often than not, been looking at Window AC units LG OLD units are almost a hundred or more cheaper, where the dual inverter types are more expensive.

 

First world problems I suppose

[ryan]

3000e cores 1p core...running at 6000w why is this a trend

[Reciever]

From a distributed computing standpoint that scenario is interesting lol

[ryan]

I have a laptop 3060 right square in the middle and all I can say is wow a 3090ti can get the same framerate as mine 4k vs 1080p

[Papusan]

https://videocardz.com/newz/gpu-price-crash-in-china-geforce-and-radeon-cards-now-sold-for-20-under-msrp-flagship-models-at-38
 

In last week alone, the flagship RTX 3090 Ti model has been sold for 9499 RMB (1415 USD), which is 38% under MSRP (14999 RMB). A bit lower, but still significant reduction applies also to RTX 3090, which can now be found for as little as 8498 RMB (1266 USD), almost 29% under MSRP.

 

Pay EVGA well above 2200$ for their 3090Ti Kingpin is crazy. This flagship graphics card will compete with next gen sub 600$ 4070(Ti) cards (coming 2022/23). Not even on 4080 performance level. Or better say slightly above a midrange card as 4070. And with the 2500$ bundle with their PSU you’ll get yesterday’s power adapter. Yep, I hate bad deals equal as bad as throwing my money down the drain for useless Trashbooks. Happy I don’t need a new card yesterday. Even then I wouldn’t jump on it. The 3090 isn’t worth a cent above 850$ as I see it now in July soon Aug (aka we still pay 50% premium above its worth for a 2 years old graphics card). And I’ll stop there. Because the 3090Ti is even a worse deal, this long into the summer. Not many days and we are into the fall and even more disgusting weather in Norway. Talked with my sons home. 7.2C outside and with nice cold flooding rain. Can’t beat disgusting? Or can we? Nvidia, AMD (and their greedy partners) is equal bad as Norwegian weather. Let them burn in with all their 3000/6000 series cards. And I hope also the miners and scalpers loose loads of money. I like it when justice prevails 🙂

 

Maybe Path tracing will help Nvidia sell 40xx series at premium (new fancy tech sell)? Let it go one year after release and then push hard as they did for RT? A lovely way to push you from 40xx series with “low Path tracing performance” to 50xx series cards. 

 

https://www.techspot.com/article/2485-path-tracing-vs-ray-tracing/

 

”Every few years, it seems like there's an amazing new technology with the promise of making games look ever more realistic. We've had shaders, tessellation, shadow mapping, ray tracing -- and now there's a new kid on the block: path tracing”

[Mr. Fox]

Well this is pretty sucky. The latest BIOS from MSI erased my memory XMP profiles and corrupted the SPD. The XMP profiles do not show in the BIOS or CPU-Z and manually overclocking the RAM is limited because it capped the voltage at 1.423V. The same condition is present on the second BIOS with older firmware and it definitely wrote/corrupted the firmware on the memory. I ordered replacement RAM. I am almost positive MSI will pull an ASUS and insist it's not their fault and refuse to take ownership of their own incompetence. I haven't contacted them yet. If the replacement memory has the same issues, then the board will need an RMA. I am almost positive it corrupted the memory firmware though. It worked flawlessly before I flashed the BIOS update for "fine tuned memory compatibility" and after rebooting it's messed up. I tested one stick at a time in each  slot and what you see on AIDA64 data is what each stick show in each slot, so the messed up data follows the module. Flashing the previous firmware changes nothing.

 

@johnksss if you haven't flashed tha firmware update, you might want to consider ignoring it.

Not only is the XMP gone on both sticks, but also look how the device description, module part number and week/year data is different on the second stick.

[johnksss]

9 minutes ago, Mr. Fox said:

Well this is pretty sucky. The latest BIOS from MSI erased my memory XMP profiles and corrupted the SPD. The XMP profiles do not show in the BIOS or CPU-Z and manually overclocking the RAM is limited because it capped the voltage at 1.423V. The same condition is present on the second BIOS with older firmware and it definitely wrote/corrupted the firmware on the memory. I ordered replacement RAM. I am almost positive MSI will pull an ASUS and insist it's not their fault and refuse to take ownership of their own incompetence. I haven't contacted them yet. If the replacement memory has the same issues, then the board will need an RMA. I am almost positive is corrupted the memory firmware though. It worked flawlessly before I flashed the BIOS update for "fine tuned memory compatibility" and after rebooting it's messed up. I tested one stick at a time in each  slot and what you see on AIDA64 data is what each stick show in each slot, so the messed up data follows the module.

 

@johnksss if you haven't flashed tha firmware update, you might want to consider ignoring it.

What is the bios number in CPUZ?

[Mr. Fox]

Just now, johnksss said:

What is the bios number in CPUZ?

A60.

[Reciever]

I remember when I was young I used to think that new models should come with better features or in general be overall a better product than the previous year.

 

These days we have measure just how much crap you have to walk through from PR events, Reviews and finally getting the product and figuring out yourself just how much of it is false.

 

I struggle to maintain a positive outlook, and that bothers me.

[Papusan]

33 minutes ago, Mr. Fox said:

Well this is pretty sucky. The latest BIOS from MSI erased my memory XMP profiles and corrupted the SPD. The XMP profiles do not show in the BIOS or CPU-Z and manually overclocking the RAM is limited because it capped the voltage at 1.423V. The same condition is present on the second BIOS with older firmware and it definitely wrote/corrupted the firmware on the memory. I ordered replacement RAM. I am almost positive MSI will pull an ASUS and insist it's not their fault and refuse to take ownership of their own incompetence. I haven't contacted them yet. If the replacement memory has the same issues, then the board will need an RMA. I am almost positive is corrupted the memory firmware though. It worked flawlessly before I flashed the BIOS update for "fine tuned memory compatibility" and after rebooting it's messed up. I tested one stick at a time in each  slot and what you see on AIDA64 data is what each stick show in each slot, so the messed up data follows the module.

 

@johnksss if you haven't flashed tha firmware update, you might want to consider ignoring it.

I wonder how DDR5 will function in next gen AMD platform. Still a huge disaster for Intel. We all are Guinea pigs for this new tech. Maybe it will be better on older platforms as times goes but I guess they’ll put more (all) work in gen 2 DDR5 platforms (Raptor lake and Intel). Good blessings to all the new coming AM5 platform buyers. But AMD won’t learn anything of the miserable things from Intel. They need to learn it their own hard ways. And it won’t be pretty l. Hail to “new tech”! One more year with sucked and screwed up tech. At least my setup worked well l before I went on summer vacation. Don’t change what’s  works! 

[johnksss]

16 minutes ago, Mr. Fox said:

A60.

Hummm, I think the highest version I had before was something like A48.

So if it messed up the memory does that mean that A60 has SPD write unlocked?

[Mr. Fox]

Just now, johnksss said:

Hummm, I think the highest version I had before was something like A48.

So if it messed up the memory does that mean that A60 has SPD write unlocked?

It would have to be unlocked, at least during the flashing process.