RTX 5060 8GB performance

[JustBenching]

The LLM issue was not relevant when the 3000 series came out, i think. This has been a problem at least with Nvidia way before anyone knew what a LLM was.
And AMD is using GDDR6, so they don't face the same issue. Nvidia could do the same.

LLMs maybe not but people were still buying high vram cards for renderstations and productivity. If you actually think about how fast nvidia cards are for non gaming workloads - it's a miracle they are as cheap as they are. I know it sucks for us gamers, but 3090 (24gb) --> 3060 (12gb) were the most sought after cards because of vram.

 

[Bomby569]

LLMs maybe not but people were still buying high vram cards for renderstations and productivity. If you actually think about how fast nvidia cards are for non gaming workloads - it's a miracle they are as cheap as they are. I know it sucks for us gamers, but 3090 (24gb) --> 3060 (12gb) were the most sought after cards because of vram.

the 3060 was very much a one off because of those HUB alarmist videos at ultra settings. Not sure it's a non gamer issue there.

 

[Red Hood]

In fact i think your example is perfect. The 5070ti is a tiny chip compared to the 9070xt and yet its faster in raster and loads faster in rt while simultaneously using less power.

I dont know why we are still arguing over this, hubs reviww shows that some games like fast vram. As the years go by, there will be more of them, not less.

NVIDIA GeForce RTX 5070 Ti Specs

NVIDIA GB203, 2452 MHz, 8960 Cores, 280 TMUs, 96 ROPs, 16384 MB GDDR7, 1750 MHz, 256 bit

www.techpowerup.com

AMD Radeon RX 9070 XT Specs

AMD Navi 48, 2970 MHz, 4096 Cores, 256 TMUs, 128 ROPs, 16384 MB GDDR6, 2518 MHz, 256 bit

www.techpowerup.com

378 mm² is tiny compared to 357 mm²?

And before you bring up the transistor count, N48 is a more dense than GB203, not bigger

 

[JustBenching]

NVIDIA GeForce RTX 5070 Ti Specs

NVIDIA GB203, 2452 MHz, 8960 Cores, 280 TMUs, 96 ROPs, 16384 MB GDDR7, 1750 MHz, 256 bit

www.techpowerup.com

View attachment 400607

AMD Radeon RX 9070 XT Specs

AMD Navi 48, 2970 MHz, 4096 Cores, 256 TMUs, 128 ROPs, 16384 MB GDDR6, 2518 MHz, 256 bit

www.techpowerup.com

View attachment 400608

378 mm² is tiny compared to 357 mm²?

And before you bring up the transistor count, N48 is a more dense than GB203, not bigger

Come on, do we have to go through this? I get the whole amd defense thing but do we really have to keep explaining the same things over and over again? Yes, im obviously referring to transistor counts, that's what dictates performance, not the die itself. Furthermore, the 5070ti isn't even a 45.6m trans count gpu since it's a cutdown GB203. Navi 48 is fricking much bigger than even a 5080 and yet it gets trumped by it. But sure, lets keep going over it again and again and again cause our love for a multibillion dollar company won't allow us to admit it. Again, im out, go have the last word, i don't care at this point, this fricking pro amd toxicity is ruining every goddamn thread. I started with a comparison between a 4060 and a 5060, both nvidia gpus, and you ended up defending amd for some ***cked up reason, lol.

EG1. LOL, I just realized why we are arguing about GDDR6 for the last 3 pages. I was like, why are we arguing about it when clearly 7 is faster, but then it hit me. The whole issue stems from the fact that a specific company only makes GDDR6 gpus and so we have to defend GDDR6 like our life depends on it! Insane...

 

[Hecate91]

Man when a small channel goes out and pay for a 5060 from his own pocket, makes me wonder about the integrity of those big and "influential" Techtubers

I don't see much integrity on a channel that just hypes up DLSS and MFG through the whole video, Nvidia probably would've sent them a gpu for that PR piece.

Well it's not really 20/50$ though. Cards with lots of vram are used for ML AI and crap. We are in fact incredibly lucky that we are only being charged 50$ for a 16gb 5060ti. I can't explain why is nvidia doing it, but it's really a gift. Just take a look at the new 9700pro, it's basically a 9070xt with more vram, and it's going to cost you like 700$ for that extra vram.

Lots of pros want to use AI locally (especially programmers), so cheap 16gb+ vram cards are godsend.

Ah yes, nvidia charging over $450 for an x60 class gpu, just because it has 16gb of vram, what a godsend. /s

The W7900 had VRAM with ECC support, the R9700 pro will probably be the same, workstation cards are always more expensive than gaming cards, but buying an x60 class card for AI makes no sense.

 

[JustBenching]

I don't see much integrity on a channel that just hypes up DLSS and MFG through the whole video, Nvidia probably would've sent them a gpu for that PR piece.

Ah yes, nvidia charging over $450 for an x60 class gpu, just because it has 16gb of vram, what a godsend. /s

The W7900 had VRAM with ECC support, the R9700 pro will probably be the same, workstation cards are always more expensive than gaming cards, but buying an x60 class card for AI makes no sense.

I know Terraware (the guy that has the channel), he has every single Ryzen CPU that ever existed, but yeah he is being paid by nvidia to promote amd cpus

So we agree, the new AI PRO 9070 will charge you 700$ for 16gb of vram. But 50$ for 8gb is ngreedia. Gotcha man, no worries.

 

[Red Hood]

Come on, do we have to go through this? I get the whole amd defense thing but do we really have to keep explaining the same things over and over again? Yes, im obviously referring to transistor counts, that's what dictates performance, not the die itself. Furthermore, the 5070ti isn't even a 45.6m trans count gpu since it's a cutdown GB203. Navi 48 is fricking much bigger than even a 5080 and yet it gets trumped by it. But sure, lets keep going over it again and again and again cause our love for a multibillion dollar company won't allow us to admit it. Again, im out, go have the last word, i don't care at this point, this fricking pro amd toxicity is ruining every goddamn thread. I started with a comparison between a 4060 and a 5060, both nvidia gpus, and you ended up defending amd for some ***cked up reason, lol.

It seems we do cause as per usual you don't read the crap that you post

The 5070ti is a tiny chip compared to the 9070xt

I'm not defending AMD here, merely pointing out you being incorrect about chip sizes which you brought up

 

[Hecate91]

I know Terraware (the guy that has the channel), he has every single Ryzen CPU that ever existed, but yeah he is being paid by nvidia to promote amd cpus

So we agree, the new AI PRO 9070 will charge you 700$ for 16gb of vram. But 50$ for 8gb is ngreedia. Gotcha man, no worries.

I've never heard of their channel, a small channel with a collection of cpu's has nothing to do with them putting up a GPU review with a bias towards the features the jacket man wants review channels to advertise for.

The R9700 Pro is a workstation card, I don't care if it's overpriced because I'd never buy one, it isn't intended for anything gaming. But the 5060 Ti is a midrange gaming card that shouldn't cost any more than $350. Like I said before I expected people to defend Nvidia for a x60 card with 16GB while crapping on AMD for the 9060XT, or even an unrelated card. AMD is selling a 16GB card for $350, for the same price Nvidia is selling an 8GB card, yet people are already ignoring it for "But the AI".

 

[LittleBro]

Come on, do we have to go through this? I get the whole amd defense thing but do we really have to keep explaining the same things over and over again? Yes, im obviously referring to transistor counts, that's what dictates performance, not the die itself. Furthermore, the 5070ti isn't even a 45.6m trans count gpu since it's a cutdown GB203. Navi 48 is fricking much bigger than even a 5080 and yet it gets trumped by it. But sure, lets keep going over it again and again and again cause our love for a multibillion dollar company won't allow us to admit it. Again, im out, go have the last word, i don't care at this point, this fricking pro amd toxicity is ruining every goddamn thread. I started with a comparison between a 4060 and a 5060, both nvidia gpus, and we ended up to defending amd for some ***cked up reason, lol.

The question is: why do you few have to go through this again, defending RTX 5060 just like you did with RTX 5060 Ti 8GB, when it's pretty clear those cards can't be recommended to anyone for that money, as they have problems. They are DOA, they struggle even in 1080p in some titles, even RTX 5060 has worse fps in few titles in 1080p than 2 years older RTX 4060. Nvidia knew about it and they tried to hide it. Would you buy RTX 5060 (Ti) 8GB for yourself? I have serious doubts. Personally, I'd recommend RTX 4060 Ti 16GB to everyone who want's to game on budget, but not RTX 5060 (Ti) 8GB. Yep, price is still to high, but at least the card is capable of handling any resolution in years to come.

 

[JustBenching]

The question is: why do you few have to go through this again, defending RTX 5060 just like you did with RTX 5060 Ti 8GB, when it's pretty clear those cards can't be recommended to anyone for that money, as they have problems. They are DOA, they struggle even in 1080p in some titles, even RTX 5060 has worse fps in few titles in 1080p than 2 years older RTX 4060. Nvidia knew about it and they tried to hide it. Would you buy RTX 5060 (Ti) 8GB for yourself? I have serious doubts. Personally, I'd recommend RTX 4060 Ti 16GB to everyone who want's to game on budget, but not RTX 5060 (Ti) 8GB. Yep, price is still to high, but at least the card is capable of handling any resolution in years to come.

Lot's of cards that are a lot more expensive than the 5060 struggle in a lot of games when trying to play them maxed out. The RTX 5060 is literally, and not by a small margin, the best value card right now. Calling it DOA because it can't max out a handful of games is ridiculous.

This is a game maxed out, at 1440p, with maxed out RT and object distance. This DOA card is twice as fast as more expensive competitors. Heck it's 50% faster than cards that cost 70% more. But yeah, totally DOA.

 

[Hecate91]

Anyways, unless its cheap as heck, id try to go with gddr7. Its only a handful of games today that take advantage of it but in the next 2 years who knows.

Having faster VRAM isn't going to help when games are still going to need more than 8GB. Nvidia could have went with GDDR6 or GDDR6X if they want to keep their margins up at their astronomical levels while still selling the card for less, an x60 class card doesn't need GDDR7 when the card is pci-e bus width limited to X8.

The 5070ti is a tiny chip compared to the 9070xt

A smaller chip which Nvidia charges the same price as AMD for. Then again, I'm not concerned with chip size when I use a GPU to play games.

People moved from RDNA3 to RDNA4 when the only thing it offers are the better RT gimmick and the better upscaling gimmick. They even sacrificed vram, since RDNA3 high end cards had more of it. I don't even know what the hell, they say one thing and then they do another

I'm sure people moved from a 4070 to a 5070 also, the only gain being the better RT and upscale gimmick, see that applies to Nvidia as well.

 

[JustBenching]

It seems we do cause as per usual you don't read the crap that you post



I'm not defending AMD here, merely pointing out you being incorrect about chip sizes which you brought up

So what word would you like me to use to describe a chip that has more transistors than another? The denser you suggested doesn't cut it, the 9060 will be denser than the 5090, but with a lot less transistors. Bigger works just fine but if we are going to pretend you didn't understand what I meant sure, suggest away, what word would you like me to use?

I'm sure people moved from a 4070 to a 5070 also, the only gain being the better RT and upscale gimmick, see that applies to Nvidia as well.

It does, but these people didn't think it's a gimmick - that's the difference

 

[Hecate91]

That's a master class lesson in simply lacking the brain power to understand what he meant.

I didn't take the lesson, "Defending Nvidia 101, how to unconditionally love the leather jacket man".
Comparing a gaming card to a workstation card makes no sense at all.

The question is: why do you few have to go through this again, defending RTX 5060 just like you did with RTX 5060 Ti 8GB, when it's pretty clear those cards can't be recommended to anyone for that money, as they have problems. They are DOA, they struggle even in 1080p in some titles, even RTX 5060 has worse fps in few titles in 1080p than 2 years older RTX 4060. Nvidia knew about it and they tried to hide it. Would you buy RTX 5060 (Ti) 8GB for yourself? I have serious doubts. Personally, I'd recommend RTX 4060 Ti 16GB to everyone who want's to game on budget, but not RTX 5060 (Ti) 8GB. Yep, price is still to high, but at least the card is capable of handling any resolution in years to come.

Because these guys couldn't possibly accept that a card from the market leader is a complete turd, Nvidia tried to hide it to the point of breaking all contact with many people in the tech press, and yet people still defend a multi-trillion dollar company for launching a mediocre card which will run out of VRAM in some games today, and in a few years might not be able to run newer titles at all. Though of course that seems to be the goal, to get consumers to line up for the next new card.

 

[Denver]

So what word would you like me to use to describe a chip that has more transistors than another? The denser you suggested doesn't cut it, the 9060 will be denser than the 5090, but with a lot less transistors. Bigger works just fine but if we are going to pretend you didn't understand what I meant sure, suggest away, what word would you like me to use?


It does, but these people didn't think it's a gimmick - that's the difference

You can argue all you want, but the 5080 is a bigger chip than the 9070XT, and has much faster memories. RDNA4 is the most bandwidth-efficient architecture at the moment.

 

[JustBenching]

Can you please stop spreading ignorance like there's no tomorrow? this isn't a kindergarten. It's obvious that you know what size means, English is your native language, isn't it?

You can argue all you want, but the 5080 is a bigger chip than the 9070XT, and has much faster memories. RDNA4 is the most bandwidth-efficient architecture at the moment.

This is from the dictionary

Big
also : large or great in quantity, number, or amount

I was referring to the quantity of transistors, so big seems to describe it just fine. GB203 has a lot less transistors than navi 48 but please, continue.

 

[AusWolf]

Come on, do we have to go through this? I get the whole amd defense thing but do we really have to keep explaining the same things over and over again? Yes, im obviously referring to transistor counts, that's what dictates performance, not the die itself. Furthermore, the 5070ti isn't even a 45.6m trans count gpu since it's a cutdown GB203. Navi 48 is fricking much bigger than even a 5080 and yet it gets trumped by it. But sure, lets keep going over it again and again and again cause our love for a multibillion dollar company won't allow us to admit it. Again, im out, go have the last word, i don't care at this point, this fricking pro amd toxicity is ruining every goddamn thread. I started with a comparison between a 4060 and a 5060, both nvidia gpus, and you ended up defending amd for some ***cked up reason, lol.

EG1. LOL, I just realized why we are arguing about GDDR6 for the last 3 pages. I was like, why are we arguing about it when clearly 7 is faster, but then it hit me. The whole issue stems from the fact that a specific company only makes GDDR6 gpus and so we have to defend GDDR6 like our life depends on it! Insane...

First, we were comparing GPUs by size. Now, by transistor count. Seriously? Who cares?

This is from the dictionary

Big
also : large or great in quantity, number, or amount

I was referring to the quantity of transistors, so big seems to describe it just fine. GB203 has a lot less transistors than navi 48 but please, continue.

I remember when squeezing more transistors into a smaller area was a good thing. Now it isn't? What changed?

 

[Kovoet]

Let's be honest according to all the reviews about this card and Nvidia trying to manipulate the media about card it is a waste of money.

 

[Hecate91]

So what word would you like me to use to describe a chip that has more transistors than another? The denser you suggested doesn't cut it, the 9060 will be denser than the 5090, but with a lot less transistors. Bigger works just fine but if we are going to pretend you didn't understand what I meant sure, suggest away, what word would you like me to use?

I think you're confusing transistor density with transistor count, I don't see how transistor count matters with two different architectures. Navi 48 is more dense with a smaller die area, and Navi 48 is competing with GB203-300, not GB203-400.

It does, but these people didn't think it's a gimmick - that's the difference

Well, no I don't have any expectation the Nvidia fans would ever call a feature or marketing they paid for a gimmick. My opinion is RT is a gimmick until midrange cards can do RT without any trickery, upscaling being the trickery needed along with fake frames in order to not take a massive hit in GPU rendered FPS.

 

[JustBenching]

Well, no I don't have any expectation the Nvidia fans would ever call a feature or marketing they paid for a gimmick. My opinion is RT is a gimmick until midrange cards can do RT without any trickery, upscaling being the trickery needed along with fake frames in order to not take a massive hit in GPU rendered FPS.

The point is - people that argued RT and upscale are both gimmicks, upgraded to RDNA 4 that only offered those gimmicks over previous gen But yeah, you keep acting like you don't understand the point, come on.

 

[the54thvoid]

How about you all go take a break?