Radeon RX Vega 64 vs. GeForce RTX 2080, Adrenalin 2019 Edition Driver Update Benchmark Test

[shankly1985]

Those are with year old drivers. Luckily they just re-did them

Brings me to my recent viewing of Vega vs 1080..

 

[Brazo]

Those are with year old drivers. Luckily they just re-did them

I am genuinely thankful! Always interesting to see the latest. Still very little in it but at least the vega now ‘bests’ the 2060 as that was getting embarrassing!

 

[TonyTurbo78]

Vega64 looking good. But its always been a fantastic card for tweakers, a perfect 1440p eye candy card for 144hz monitors.

 

[melmac]

AIB partners do not sell reference cards.

Any 2080 from say Asus is nothing like any 2080 from say EVGA.

The only reference card out there is the FE one as it is the only one NVidia sell themselves.

As to overclocked concerning NVidia FE cards go and checkout all the FE TU102 cards, according to NVidia own slides the slowest one is overclocked and the fastest one is not lol.

I am pretty sure this is a wind up now, because there is no way you actually believe what you are saying.

Just in case you actually do think what you are saying is right. We are talking about the 2080 card here not sure why you keep bringing the Titan and TU102 into the conversation. (Even though the 2080Ti FE is the same as the 2080, it's overclocked, not reference spec)

Nvidia's founder edition 2080's are not reference spec cards.

Here is a quote from Anandtech.

And on that note, the ‘reference’ Founders Edition models are no longer reference; the GeForce RTX 2080 Ti, 2080, and 2070 Founders Editions feature 90MHz factory overclocks and 10W higher TDP, and NVIDIA does not plan to productize a reference card themselves.

The reference specs for the 2080 are listed on Nvidia's site, I posted a link to them. Those are the baseline specs for AIB cards.

You are saying the Palit card is only a 15hz OC because the FE card is 1800Mhz. LOL It's not based on the Founders Edition clocks it's based on the Reference spec clocks.

But lets say by some wild stretch of imagination that you are right and every review and Nvidia are wrong. How does it explain the Gigabyte Windforce OC edition? Its boost is running at 1785. What's that a -15hz OC. hahahaha yes indeed. Or the MSI Ventus OC editon running at 1800MHz, a 0 increase. You think manufacturers would be putting "OC" on their cards with minus values for an overclock?

Or maybe its simply that you are wrong. The Windforce is 75hz overclock because its baseline is the 1710Mhz of the Reference spec not the Overclocked 1800Mhz of the Founder's edition.

Lets do a simple sum up.

You are saying that Palit card running at 1815 is only a 15hz overclock based on the FE2080 specs, which you say is a reference card. Your evidence of this is that Nvidia don't sell any Reference cards with the 1710Mhz clock speeds. Oh, and you think it's all marketing.

I am saying that the 1815Mhz is actually a 105hz overclock because its based on the Reference spec clocks not the FE card clocks because the FE card is overclocked, It's not reference. My evidence for this
1. Nvidia list the reference specs for the 2080 on their site. The clocks are 1710.
2. Nvidia state on their site that the FE cards are Overclocked by 90Hz.
3. Any review of the 2080 says that the FE cards are overclocked.
4. If the FE specs were actually the reference specs then there would be some card companies guilty of false advertising as some of the so called Overclocked cards are less than the 1800Mhz of the FE edition.
5. Several hardware sites, like the quote from the Anandtech site above, have stated that Nvidia have no plans to make a board based on the Reference specs.

 

[Kaapstad]

I am pretty sure this is a wind up now, because there is no way you actually believe what you are saying.

Just in case you actually do think what you are saying is right. We are talking about the 2080 card here not sure why you keep bringing the Titan and TU102 into the conversation. (Even though the 2080Ti FE is the same as the 2080, it's overclocked, not reference spec)

Nvidia's founder edition 2080's are not reference spec cards.

Here is a quote from Anandtech.



The reference specs for the 2080 are listed on Nvidia's site, I posted a link to them. Those are the baseline specs for AIB cards.

You are saying the Palit card is only a 15hz OC because the FE card is 1800Mhz. LOL It's not based on the Founders Edition clocks it's based on the Reference spec clocks.

But lets say by some wild stretch of imagination that you are right and every review and Nvidia are wrong. How does it explain the Gigabyte Windforce OC edition? Its boost is running at 1785. What's that a -15hz OC. hahahaha yes indeed. Or the MSI Ventus OC editon running at 1800MHz, a 0 increase. You think manufacturers would be putting "OC" on their cards with minus values for an overclock?

Or maybe its simply that you are wrong. The Windforce is 75hz overclock because its baseline is the 1710Mhz of the Reference spec not the Overclocked 1800Mhz of the Founder's edition.

Lets do a simple sum up.

You are saying that Palit card running at 1815 is only a 15hz overclock based on the FE2080 specs, which you say is a reference card. Your evidence of this is that Nvidia don't sell any Reference cards with the 1710Mhz clock speeds. Oh, and you think it's all marketing.

I am saying that the 1815Mhz is actually a 105hz overclock because its based on the Reference spec clocks not the FE card clocks because the FE card is overclocked, It's not reference. My evidence for this
1. Nvidia list the reference specs for the 2080 on their site. The clocks are 1710.
2. Nvidia state on their site that the FE cards are Overclocked by 90Hz.
3. Any review of the 2080 says that the FE cards are overclocked.
4. If the FE specs were actually the reference specs then there would be some card companies guilty of false advertising as some of the so called Overclocked cards are less than the 1800Mhz of the FE edition.
5. Several hardware sites, like the quote from the Anandtech site above, have stated that Nvidia have no plans to make a board based on the Reference specs.

I keep mentioning the TU102 chip based cards for a reason.

Both the 2080 Ti and RTX Titan both use the chip.

The Titan brand always uses the reference cooler which is the same type found on all the other FE cards.

Check out the specs for the 2080 Ti and you will notice NVidia have the FE one down as overclocked @1635mhz boost on the core.





Now check out the specs for the RTX Titan and you can see the core runs @1770mhz boost and it is not overclocked.






How can the Titan be a reference card that always uses the reference cooler and the FE 2080 Ti not be a reference card?

How can the FE 2080 Ti be overclocked and run slower than a fully enabled RTX Titan which is not overclocked?

 

[layte]

Those are with year old drivers. Luckily they just re-did them

The rest of the omitted results can be found here for anyone interested. https://www.techpowerup.com/reviews/Palit/GeForce_RTX_2070_GameRock_Premium/27.html

 

[melmac]

I keep mentioning the TU102 chip based cards for a reason.

Both the 2080 Ti and RTX Titan both use the chip.

The Titan brand always uses the reference cooler which is the same type found on all the other FE cards.

Check out the specs for the 2080 Ti and you will notice NVidia have the FE one down as overclocked @1635mhz boost on the core.





Now check out the specs for the RTX Titan and you can see the core runs @1770mhz boost and it is not overclocked.






How can the Titan be a reference card that always uses the reference cooler and the FE 2080 Ti not be a reference card?

How can the FE 2080 Ti be overclocked and run slower than a fully enabled RTX Titan which is not overclocked?

In your very post you see the 2080Ti reference Specs listed right beside the Founders Edition Specs. You can plainly see the Founders Edition cards are running at a 90Hz higher boost clock. Now go on the Retail part of this website and look up the specs of the AIB cards. You will see that they start at the Reference Specs not the Founders edition specs.

Now check out OC versions of the Various cards, you will see that they are been Overclocked based on the reference specs not the founders edition specs.

The Titan is overclocked. It's running at 1770Mhz, The TU102 baseline spec is 1545hz boost. Not quite sure what's so hard to understand here. It's just not called a overclock on the Titan card because there is only one, there aren't going to be any variants or AIB versions. It's just the RTX Titan card.

The uniqueness of the Titan card and the way they brand it does not change anything in our discussion though. For the 2080ti, 2080 and 2070 the founders editions are all overclocked, it even says it in the specs. The AIB card Overclocks are based on those reference specs listed on the Nvidia's site. For example, compare the reference specs for the 2080Ti that you have listed above and the specs of Msi Ventus which has the same clock speeds(1545Mhz), now compare them both to the MSI Ventus OC edition which has clock speeds of 1635Mhz.

So the Palit 2080 used in the Review has a 105hz overclock not a 15Hz one.

 

[Kaapstad]

In your very post you see the 2080Ti reference Specs listed right beside the Founders Edition Specs. You can plainly see the Founders Edition cards are running at a 90Hz higher boost clock. Now go on the Retail part of this website and look up the specs of the AIB cards. You will see that they start at the Reference Specs not the Founders edition specs.

Now check out OC versions of the Various cards, you will see that they are been Overclocked based on the reference specs not the founders edition specs.

The Titan is overclocked. It's running at 1770Mhz, The TU102 baseline spec is 1545hz boost. Not quite sure what's so hard to understand here. It's just not called a overclock on the Titan card because there is only one, there aren't going to be any variants or AIB versions. It's just the RTX Titan card.

The uniqueness of the Titan card and the way they brand it does not change anything in our discussion though. For the 2080ti, 2080 and 2070 the founders editions are all overclocked, it even says it in the specs. The AIB card Overclocks are based on those reference specs listed on the Nvidia's site. For example, compare the reference specs for the 2080Ti that you have listed above and the specs of Msi Ventus which has the same clock speeds(1545Mhz), now compare them both to the MSI Ventus OC edition which has clock speeds of 1635Mhz.

So the Palit 2080 used in the Review has a 105hz overclock not a 15Hz one.

The Titan is not overclocked as there are no other SKUs of it and so it is the reference card.

NVidia set the default clockspeed for the FE cards, for marketing they can call them overclocked underclocked over the moon or anything else but until they sell any other SKU apart from FE cards they are the reference cards.

As to the Titan why has it got more overclocking headroom than a Vega 64 card if it is overclocked, answer because it is not.

Get over it the review in the OP was totally messed up and all you are doing is arguing points of order.

You can only have an overclocked card if there is a baseline one using exactly the same chip and cooler that is not overclocked otherwise it is the default reference card.

 

[james.miller]

Nvidia dont make a reference 2080ti. They have a reference design, and the FE isnt it. nVidia don't refer to the FE as a reference card anywhere. That should be good enough for anybody.

You can only have an overclocked card if there is a baseline one using exactly the same chip and cooler that is not overclocked otherwise it is the default reference card.

The same chip and cooler? How often to AIB cards even use the same cooler as a reference card (when a reference card even exists...) ?

 

[Gibbo]

The Titan is not overclocked as there are no other SKUs of it and so it is the reference card.

NVidia set the default clockspeed for the FE cards, for marketing they can call them overclocked underclocked over the moon or anything else but until they sell any other SKU apart from FE cards they are the reference cards.

As to the Titan why has it got more overclocking headroom than a Vega 64 card if it is overclocked, answer because it is not.

Get over it the review in the OP was totally messed up and all you are doing is arguing points of order.

You can only have an overclocked card if there is a baseline one using exactly the same chip and cooler that is not overclocked otherwise it is the default reference card.

FE cards are overclocked, they use the different board+GPU, this board+GPU cost more which AIB's buy for their OC models. The base clocks or reference clocks whichever people prefer is the cheaper board+GPU options, such as Palit Dual, these have the stock / reference clocks.

 

[subbytna]

I don't understand why you all get so riled up about random no name videos with obviously dodgy testing methodology. Surely it would be better to look, scratch your chin and move on to something else.

If people want to take it as given, so what, let them.

Over 10 years in the OCUK forums and you still wonder this??

It's the creatures natural trolling habitat

 

[melmac]

The Titan is not overclocked as there are no other SKUs of it and so it is the reference card.

NVidia set the default clockspeed for the FE cards, for marketing they can call them overclocked underclocked over the moon or anything else but until they sell any other SKU apart from FE cards they are the reference cards.

As to the Titan why has it got more overclocking headroom than a Vega 64 card if it is overclocked, answer because it is not.

Get over it the review in the OP was totally messed up and all you are doing is arguing points of order.

You can only have an overclocked card if there is a baseline one using exactly the same chip and cooler that is not overclocked otherwise it is the default reference card.

No, you get over it, You are wrong. Plain and simple. I will quote this again from Anandtech and highlight the important parts.

And on that note, the ‘reference’ Founders Edition models are no longer reference; the GeForce RTX 2080 Ti, 2080, and 2070 Founders Editions feature 90MHz factory overclocks and 10W higher TDP, and NVIDIA does not plan to productize a reference card themselves.

Read the posts from Gibbo or James.Miller

The 2080 FE editon is not a reference card.

The Pailit card in the review had a 105Hz overclock not a 15Hz overclock as you stated.
The Vega card had the same small overclock, not a Big overclock as you tried to make it out to be.

Arguing points of Order, no, I am telling you that you are wrong, you just aren't man enough to accept you are wrong.

And just on the Titan card and your comparison to the overclocking headroom on the Vega 64. WUT??? How does the overclocking headroom on a completely different process prove that the Titan isn't overclocked? Look at the reference specs of the 2080ti. 1350Mhz base clock, 1545Mhz boost. The Titan is the same chip just fully unlocked, and it's based on the same specs except it's overclocked like the Founders Edition. It has 1350Mhz base clock and 1770Mhz boost. The 2080ti FE has 1350 base and 1635Mhz boost.

Just to help you out, 1770Mhz is higher than 1545Mhz.

 

[melmac]

FE cards are overclocked, they use the different board+GPU, this board+GPU cost more which AIB's buy for their OC models. The base clocks or reference clocks whichever people prefer is the cheaper board+GPU options, such as Palit Dual, these have the stock / reference clocks.

Thank you!! But, I don't think he will believe you.

 

[melmac]

Nvidia dont make a reference 2080ti. They have a reference design, and the FE isnt it. nVidia don't refer to the FE as a reference card anywhere. That should be good enough for anybody.



The same chip and cooler? How often to AIB cards even use the same cooler as a reference card (when a reference card even exists...) ?

Basically this. But, I think if you have been following along you will see that apparently it's not enough.

 

[Gibbo]

Thank you!! But, I don't think he will believe you.

Well NVIDIA confirmed as much at release as I questioned why FE cards had more RTX-0P cores, thinking there was a hardware difference.
They confirmed there was no hardware difference, they just use a higher binned GPU (more expensive board+GPU as found on AIB OC cards) and that RTX-OP cores is directly linked to core speed, so more core speed the more RTX-OPS they can advertise. Basically just marketing.

But yes FE cards are indeed overclocked.

 

[layte]

Over 10 years in the OCUK forums and you still wonder this??

It's the creatures natural trolling habitat

It just boggles the mind that it is the same people who have been at it for years. I believe this famous quote covers it.

"Insanity is doing the same thing over and over again and expecting different results."

 

[Poneros]

I am genuinely thankful! Always interesting to see the latest. Still very little in it but at least the vega now ‘bests’ the 2060 as that was getting embarrassing!

Also worth keeping in mind that those are reference models, which no one sells anymore. In reality the performance for Vegas you can buy are still higher, as comparable or lower prices to the Nvidia counterparts, at least in the UK.
Still don't trust TPU for accurate numbers either way, they always had shoddy methodology. The best reviews come from "foreign" (non-English speaking) sources (Tom's Hardware Germany, etc).

 

[Brazo]

Also worth keeping in mind that those are reference models, which no one sells anymore. In reality the performance for Vegas you can buy are still higher, as comparable or lower prices to the Nvidia counterparts, at least in the UK.
Still don't trust TPU for accurate numbers either way, they always had shoddy methodology. The best reviews come from "foreign" (non-English speaking) sources (Tom's Hardware Germany, etc).

I think it’s also worth bearing in mind that vega at 100% fan speed is not indicative of true performance as it’s simply unplayable so I will take ‘suicide run’ performance figures with a pinch of salt too.

 

[Kaapstad]

Well NVIDIA confirmed as much at release as I questioned why FE cards had more RTX-0P cores, thinking there was a hardware difference.
They confirmed there was no hardware difference, they just use a higher binned GPU (more expensive board+GPU as found on AIB OC cards) and that RTX-OP cores is directly linked to core speed, so more core speed the more RTX-OPS they can advertise. Basically just marketing.

But yes FE cards are indeed overclocked.

This is my point hardware wise they not the same as the lower binned chips and boards.

 

[Tom B]

Those are with year old drivers. Luckily they just re-did them

While that's great for the Vega 64, this doesn't look great for the VII to be honest. It's supposedly ~30% better than the Vega 64. 83*1.3=107.9, that puts it level with the 1080 Ti and 10% behind the 2080 that it's supposed to compete with.