Is the fastest GPU ALWAYS the best?

[BaidDSB]

Whats up with your Cyberpunk numbers?

 

its bigger than EVERYONE ELSE's.  You sure you didnt keep DLSS 3 on?

 

Maybe you triple checked the sponsors instead of the settings

[Alvarojsrf]

Can I have the source from the 8:50 charts, please?

[Slopokdave]

The CP2077 4K RT On, DLSS Off chart is wrong. Either DLSS was on or it's actually 1440p. 

 

Your numbers are double of everyone else's with those same settings.

[powertoys]

I'm going to spend $1600 on a 4090 so I can future-proof my system!

 

* ... 2 years later ... *

 

I'm going to spend $1800 on a 5090 so I can future-proof my system!

[BobVonBob]

I can't get over how janky the "RTX 4090" mark on this card looks. It's like Nvidia accidentally turned off bold for the "4090".

 

[Tony1337]

Paid off to fudge the CP2077 numbers?

[Middcore]

42 minutes ago, BobVonBob said:

I can't get over how janky the "RTX 4090" mark on this card looks. It's like Nvidia accidentally turned off bold for the "4090".

 

 

It looks like something you'd see on a fake card. 

[Deadpool2onBlu-Ray]

5 minutes ago, Middcore said:

 

It looks like something you'd see on a fake card. 

Yes, I hate the new font. Although the cooling on the new founders looks excellent. I really want a 4080 16gb if they ever get down to "4080" 12gb prices

[BaidDSB]

 

[dogwitch]

2 hours ago, Slopokdave said:

The CP2077 4K RT On, DLSS Off chart is wrong. Either DLSS was on or it's actually 1440p. 

 

Your numbers are double of everyone else's with those same settings.

i really wonder why nvidia push 2077 so hard in press and video.

[MultiGamerClub]

22 minutes ago, dogwitch said:

i really wonder why nvidia push 2077 so hard in press and video.

unoptimized game that is in a barely playable state on 3000-series cards, even 3090's..

 

Probably just marketing to show how better it runs.. Without the RTX enabled with a 3080TI the game runs fine maxed out with 120-150fps, but with RTX enabled its pretty bad.. 50-60 ish, some times not even that.

[Bluon]

1 hour ago, BaidDSB said:

 

Will be interesting to find out what caused it. I really can't see this as a simple case of the wrong option being switched on/off.

[BaidDSB]

@LinusTechfigured out the problem with the cyberpunk numbers yet?

[Greyspectre]

Am I just having a selective memory were there significantly less price complaints about the 3090 (non-ti) than the 4090? I remember price complaints about the 2080ti and when the Titan series were initially released, I don't remember price being such a big focus on reviews of the 3090. Per my memory the response regarding price on the 4090 has been crazy compared to the 3090 when the release prices for the two cards are pretty comparable.

[BobVonBob]

1 hour ago, Greyspectre said:

Am I just having a selective memory were there significantly less price complaints about the 3090 (non-ti) than the 4090? I remember price complaints about the 2080ti and when the Titan series were initially released, I don't remember price being such a big focus on reviews of the 3090. Per my memory the response regarding price on the 4090 has been crazy compared to the 3090 when the release prices for the two cards are pretty comparable.

Honestly I don't think many people really care about the price of the 4090, the Titans and XX90s have always been for people with an abundance of money or a lack of spending discipline. I think it's more that prices are on people's minds more now that we've been through the madness of the last two years, and Nvidia really tightened the screws on the "achievable" tiers.

 

At the time of release people thought the 3090 was the card for nutters, and the 3080 was expensive, but just affordable enough and powerful enough for people to stomach it. Then the prices of everything jumped 200%, but people were still buying cards, including the card for nutters, and suddenly Nvidia realized how much free money was roaming around in the pockets of computer enthusiasts that they could just have by putting a bigger number on the sticker.

 

Now we're seeing Nvidia double down on the pricing, especially in the lower tiers with the 4080 going up $500 USD (70%) and the 4070 (/"4080 12 GB") going up $400 (80%) compared to the 3080 and 3070. I think that's really where the pricing unrest is coming from.

[Ultraforce]

3 hours ago, BaidDSB said:

@LinusTechfigured out the problem with the cyberpunk numbers yet?

They found the cause FFX upscaling was forcing on even though it was sent to off. Testing with all cards has been rerun and the video will be updated.

[Uttamattamakin]

5 hours ago, MultiGamerClub said:

unoptimized game that is in a barely playable state on 3000-series cards, even 3090's..

 

Probably just marketing to show how better it runs.. Without the RTX enabled with a 3080TI the game runs fine maxed out with 120-150fps, but with RTX enabled its pretty bad.. 50-60 ish, some times not even that.

I have Played it on 4K 30 FPS with RTX on a 3080, granted there are BIG drops.  1440p same settings otherwise it is fine.  I really don't know what is wrong with other peoples computers. 

[Spotty]

@GabenJr if you're updating the video to fix the bad benchmarks can you also fix where you call the 12VHPWR cable a "ATX 3.0 cable". Calling it an ATX 3.0 cable would be like calling a SATA power cable an "ATX 2.0 cable". Some ATX 3.0 power supplies that meet certain requirements will be required to include the 12VHPWR cable, but not all ATX 3.0 power supplies require it and a power supply does not need to be ATX 3.0 to have it. For example the Gigabyte power supply short Circuit recently did a sponsored video for has the 12VHPWR cable and is not an ATX 3.0 power supply and fails to meet ATX 3.0 requirements, even though short Circuit falsely advertised it as an ATX 3.0 power supply (which I still don't know if was a mistake on LMG's part or intentional false advertisement by Gigabyte who sponsored the video). Calling it an ATX 3.0 cable causes people to be confused thinking you need an ATX 3.0 power supply to use the cable or thinking that any PSU with the cable must be ATX 3.0.

[Ultraforce]

Here's the comment from the youtube post that LTT posted displaying that info 4 hours ago.

CORRECTION: We're working on updating the video, but in the meantime, our numbers for Cyberpunk 2077 were with FidelityFX Upscaling enabled. We specifically didn't have this enabled, but stability issues with the bench seems to have messed with the settings. We've re-run all of the numbers for each card:
*No RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 54, 69, 81
- RTX 3090 Ti: 43, 46, 56
- RTX 3090: 35, 43, 50
- RX 6950 XT: 30, 39, 46
*RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 36, 39, 44
- RTX 3090 Ti: 17, 22, 26
- RTX 3090: 16, 19, 23
- RX 6950 XT: 10, 11, 13
*RT + DLSS (1%low, 5%low, avg):
- RTX 4090: 94, 97, 108
- RTX 3090 Ti: 58, 60, 67
- RTX 3090: 52, 53, 61
- RX 6950 XT: N/A
-AY

[BaidDSB]

9 minutes ago, Ultraforce said:

Here's the comment from the youtube post that LTT posted displaying that info 4 hours ago.

CORRECTION: We're working on updating the video, but in the meantime, our numbers for Cyberpunk 2077 were with FidelityFX Upscaling enabled. We specifically didn't have this enabled, but stability issues with the bench seems to have messed with the settings. We've re-run all of the numbers for each card:
*No RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 54, 69, 81
- RTX 3090 Ti: 43, 46, 56
- RTX 3090: 35, 43, 50
- RX 6950 XT: 30, 39, 46
*RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 36, 39, 44
- RTX 3090 Ti: 17, 22, 26
- RTX 3090: 16, 19, 23
- RX 6950 XT: 10, 11, 13
*RT + DLSS (1%low, 5%low, avg):
- RTX 4090: 94, 97, 108
- RTX 3090 Ti: 58, 60, 67
- RTX 3090: 52, 53, 61
- RX 6950 XT: N/A
-AY

this is 4K right? no point doing 1080p benchmarks on these cards

[Spotty]

24 minutes ago, Ultraforce said:

Here's the comment from the youtube post that LTT posted displaying that info 4 hours ago.

CORRECTION: We're working on updating the video, but in the meantime, our numbers for Cyberpunk 2077 were with FidelityFX Upscaling enabled. We specifically didn't have this enabled, but stability issues with the bench seems to have messed with the settings. We've re-run all of the numbers for each card:
*No RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 54, 69, 81
- RTX 3090 Ti: 43, 46, 56
- RTX 3090: 35, 43, 50
- RX 6950 XT: 30, 39, 46
*RT, no DLSS (1%low, 5%low, avg):
- RTX 4090: 36, 39, 44
- RTX 3090 Ti: 17, 22, 26
- RTX 3090: 16, 19, 23
- RX 6950 XT: 10, 11, 13
*RT + DLSS (1%low, 5%low, avg):
- RTX 4090: 94, 97, 108
- RTX 3090 Ti: 58, 60, 67
- RTX 3090: 52, 53, 61
- RX 6950 XT: N/A
-AY

 

Quite a bit of difference in the updated 6950XT results as well. The new results see the 6950XT go from average 28 FPS to 46 FPS No RT / No DLSS.  Why is there now a 65% improvement on the 6950XT results over what they originally got?

 

The updated results seem to be in line with what other reviews show, just not sure why the original result for the 6950XT was much lower.

[dogwitch]

16 minutes ago, Spotty said:

 

Quite a bit of difference in the updated 6950XT results as well. The new results see the 6950XT go from average 28 FPS to 46 FPS No RT / No DLSS.  Why is there now a 65% improvement on the 6950XT results over what they originally got?

 

The updated results seem to be in line with what other reviews show, just not sure why the original result for the 6950XT was much lower.

driver bug issue. i think i know what they ran into thru.

[Zupadupa]

At first I didn't watch this video because the title was too click-baity and did not mention 4090, and thumbnails on youtube are pretty small so I couldn't read 4090 there too.

 

Sometimes plastering 4090 in the title works best!

 

Great video however, especially the chapters talking about DisplayPort + PCI Express that not every reviewer seemed to mention - Nvidia really dropped the ball there. And only 1 HDMI!

[Dogzilla07]

53 minutes ago, Zupadupa said:

Nvidia really dropped the ball there.

Maybe not, from the aspect of the company/shareholders. Lack of DP 2.0 means buyers have to get a 5000 series card for 4K >120Hz. It mitigates the severity of another 1080 Ti situation.

[Zodiark1593]

Maybe a hot take, but I don't think the power consumption is a big deal with this card. This card is seemingly aimed at the people that would've formerly be running SLI, which itself draws a lot of power out of virtue of literally running two cards. So increasing the performance and power ceiling makes sense. So long as the performance scales accordingly, raising the power ceiling is fine, and should represent a minimal hit to performance/watt. If efficiency is similar or better than the outgoing generation, I see little reason to raise objection based on power consumption. Either buy it if you need the performance and can accommodate for the requirements, or don't, and pick something that you can accommodate.

 

I'm not the target market, so this card is pretty irrelevant to me. I'm more of a $200 kind of customer (which looks absolutely barren for options that are appreciably better than my 7-year old GTX 960, also bought at the same price).