Vega FE Hybrid Mod - Passing the 1080 but 400W+ Power Draw

[bomerr]

9 minutes ago, leadeater said:

You even then started to try and use what I was pointing out in regards to Quadro and GeForce. You're stance on how to compare products is flip flopping all over the place.

 

My original post and all others after it had nothing to do with Vega FE at all, the whole point was how you can not use one product line to draw any conclusive statements about another different line of products i.e. GeForce and Quadro which half way through you started to take that stance. You can not also use the difference between GeForce and Quadro to mean the same difference will apply to AMD product lines, that shouldn't be done.

 

So do you agree or not, can you or can you not draw conclusive evidence from a product by testing another different product line?

You can draw strong (but not perfect) conclusion by comparing GeForce to Quadro and Readon to FirePro. It's the exact same situation as comparing Xeons to Core iX CPUs. All these companies are essentially developing 1 new architecture and then slightly modifying it for different price performance targets and price brackets. So we can infer the core performance of the GPU. 

 

It's also worth mentioning that in this case Vega isn't even a FirePro card, so that leads even stronger credence that RX Vega will perform similar to Vega FE. 

[JediFragger]

On 7/14/2017 at 0:09 AM, Sierra Fox said:

It more sounds like they did testing, found its performance was close enough to their intended target and said "**** it"

Fixed for truth

[leadeater]

 

30 minutes ago, bomerr said:

You can draw strong (but not perfect) conclusion by comparing GeForce to Quadro and Readon to FirePro. It's the exact same situation as comparing Xeons to Core iX CPUs. All these companies are essentially developing 1 new architecture and then slightly modifying it for different price performance targets and price brackets. So we can infer the core performance of the GPU. 

 

It's also worth mentioning that in this case Vega isn't even a FirePro card, so that leads even stronger credence that RX Vega will perform similar to Vega FE. 

 

True you can draw some rather good information from tests like that, it's not going to be totally accurate but it's worth investigating which is why some reviewers brought their own cards since no review samples were being given out. You just have to be careful since you are not looking at the actual product you will be buying, you then have to take a bit more caution when it is shown that potentially not all architecture features are running/functioning on the card that has been tested.

 

Then there is the other factors which relates to the above, Vega FE is running different firmware and uses a different driver fork, potentially two different ones. I'm not sure on the real differences between FirePro, Radeon and Vega FE drivers and Vega FE's game mode and workstation mode.

 

It's comments like these that I object to, they are either making statements about a product that isn't even out yet as if they are true or are just inflammatory and damaging to the conversation and almost always derail the current conversation chain in to an utter mess. See in spoiler.

 

Spoiler

On 7/14/2017 at 11:25 PM, Alexokan said:

I can't get behind this. EVEN IF RX Vega performs on par with the 1080ti - it will consume twice as much power (maybe more). 

On 7/14/2017 at 9:52 PM, LAwLz said:

I love how "just wait for future updates" is the default answer whenever an AMD product doesn't perform as well as expected.

 

Vega uses a ton of power and doesn't perform that well? Just wait for a driver update and it will totally be fine. Trust me, because nobody can disprove me. 

 

Ryzen doesn't perform well in certain applications, and has a ton of issues? Just wait for future updates. 

On 7/15/2017 at 5:47 AM, AnonymousGuy said:

Just wait for optimizations to come after launch.

 

...am I doing "typical AMD fan" right?

On 7/15/2017 at 8:01 AM, LAwLz said:

Just do what all the AMD fanboys are doing and say that the card is amazing and future driver updates will make it better than Nvidia.

It is impossible to disprove so you can make the claim however much you want, nobody can argue against it.

 

I can also find many more examples of these types of comments and how they poison topics, sometimes they don't but usually they do. There's a difference between saying "Looking at these results it's rather worrying for the gaming cards how much power they will use" and "RX Vega will consume twice as much power".

[Belgarathian]

On 7/14/2017 at 9:50 PM, Valentyn said:

-snip-

GOD DAMN, that's a sexy looking card. 

[LAwLz]

11 hours ago, leadeater said:

-snip-

I think this conversation is over. We are talking about two different things and therefore it is not leading anywhere. Whenever I bring up my point (such as "there is no such thing as reducing workstation performance to increase gaming performance, or vice versa") you ignore it and go on to talk about other things such as how Quadro cards might perform better in some tasks, which is something I have never disagreed with.

 

You say I dismiss evidence, but the problem is that not even the evidence you posted agrees with you (even the author of the article you linked agrees with me, and the test was missed crucial things as well such as loading the appropriate profile).

A Quadro card performing better than a GeForce card does not prove that a GeForce card will perform better than a Quadro card. That comparison is a one way street.

 

Feel free to comment again if you want, but I won't respond since it will just go around in circles again.

 

 

10 hours ago, leadeater said:

It's comments like these that I object to, they are either making statements about a product that isn't even out yet as if they are true or are just inflammatory and damaging to the conversation and almost always derail the current conversation chain in to an utter mess. See in spoiler.

The statements I made are right though.

"Future drivers will fix this" is unfalsifiable. That statement is not based on facts, and I think people who are arguing for or against a product without any evidence to back them up (and yes, Vega Fe is evidence for how Vega RX will perform. It's not 100% accurate but it is perfectly acceptable to draw conclusions based on the results) are kind of silly.

 

That's the difference between me (because I assume you are including me in the "obnoxious anti-Vega hater" group) and the AMD fanboys is that I actually have some evidence to back up my stance with, while they do not.

My evidence might not be 100% accurate but nothing ever will be because of things like differences in ASIC quality. Benchmarks done on one Vega RX will not be 100% accurate representation of what a different Vega RX will be, so dismissing all benchmarks just because "it's not the exact same product" is silly. Using benchmarks from similar products is totally fine and OK to do to draw conclusions.

Don't dismiss someone for using let's say 1700 benchmarks to gauge the performance of a 1700X.

[mr moose]

11 hours ago, leadeater said:

 

There's a difference between saying "Looking at these results it's rather worrying for the gaming cards how much power they will use" and "RX Vega will consume twice as much power".

 

And therein lies the difference between an enthusiast and a professional.

[leadeater]

9 hours ago, LAwLz said:

I think this conversation is over. We are talking about two different things and therefore it is not leading anywhere. Whenever I bring up my point (such as "there is no such thing as reducing workstation performance to increase gaming performance, or vice versa") you ignore it and go on to talk about other things such as how Quadro cards might perform better in some tasks, which is something I have never disagreed with.

Yes I agree that it is over but that wasn't EVER what I was discussing, that is what you were trying to discuss which brings me back to you still don't have any idea at all what I was saying.

 

From your first reply to me you didn't get the point and continued to try and push a performance conversation which I had to respond to, and continue to try and point out during that discussion performance was not the point.

 

Please actually go back and try and understand the point, it very much applies to you.

 

On 7/15/2017 at 0:04 PM, leadeater said:

I can say with pretty good confidence that a few things would have happened, many would say that it's not a Quadro and doesn't have optimized drivers, , and that many features are not available on that card which is applicable to those use cases.

You did exactly this then tried to push your point by arguing to dirty best practice in benchmarking by making a software hack to enable a feature a requirement of said testing which was not required to get Solidworks to run on GeForce but one feature of Solidworks (RealView), a point you missed or ignored. Forcing that feature on allowed Puget to run the complete suite of tests on both Quadro and GeForce to show the differences.

 

You then doubled down on bad benchmarking practices by making loading an unsupported Quadro application profile on to GeForce cards a requirement, without then even providing any evidence of the performance increase claimed. If you had actually bothered to look online for people trying to force enable RealView and also apply the performance tweak you would have seen how it doesn't always work and many say it does not increase performance it only enables RealView

 

Quote

The result? This enabled Realview in SolidWorks for the GTX 780. It makes things look pretty, which of course is nice, but there is no performance improvement. Rotations are not smoother, and if Edges are displayed performance tanks, just like it will with any Geforce card.
So this trick helps gets Realview enabled, but it will not help performance. You won't get Quadro performance. You do get something for free though. 

https://hardforum.com/threads/looking-at-a-workstation-gpu-like-a-quadro-advice.1925719/

 

Puget never stated what they did to enable RealView, for all you know that is what they did.

 

Then after the tests I gave you shows that not even a clearly superior GPU based card was able to match a basic Quadro card you demanded I provide one that is the same so I obliged followed by you not even reading the review properly and missing the fact there was a GeForce and Quadro card of equivalent GPU included.

 

You then moved on to now requiring something even worse be done than all the terrible practices before that by saying that GeForce and Quadro use different firmware so using Quadro drivers on a GeForce card is required, or was it GeForce drivers on to a Quadro? Really load unsupported drivers?

 

GeForce is GeForce

Quadro is Quadro

 

If you want to compare any facet of each against each other you have to do so under strict testing conditions using the hardware as designed if you want your results to be credible.

 

Then to top all that off you didn't actually notice the tests were on spec perfview not the applications you mentioned, there are tests in spec perfview which simulate the demand of those types of applications.

 

https://www.spec.org/gwpg/gpc.static/vp12info.html

 

How can I take what you are saying seriously when you are making the same sliding continual excuses of "fanboys" constantly shifting the requirements and ignoring an independent test who concluded always use Quadro or a test using the industry standard for testing professional graphics performance.

 

To say you played out my scenario is an understatement and it was never intended to be a prediction but something to think about before posting any complaints about AMD "fanboys" because you know darn well if that had been a real life event this same thing would have played out but for Nvidia Pascal and not Vega.

 

You have also been saying you have not been making any conclusive statements about gaming Vega cards or presenting things as if they were facts about them but you have and if you really want I can show them to you. Maybe you just don't realize you have been doing it.

[Jito463]

42 minutes ago, LAwLz said:

The statements I made are right though.

"Future drivers will fix this" is unfalsifiable. That statement is not based on facts

That statement may not be based on facts (because, at this moment, there are no "facts" surrounding RX Vega, only speculation), but that doesn't mean we can't have some idea of whether that's true or not.  We know from past experience that AMD has a history of releasing drivers on day one that are improved upon (sometimes considerably) with future updates.  That's not speculation, that is a fact.  I'm not arguing how much performance will be gained, but merely that we know performance can improve from driver updates.

 

44 minutes ago, LAwLz said:

(and yes, Vega Fe is evidence for how Vega RX will perform. It's not 100% accurate but it is perfectly acceptable to draw conclusions based on the results)

Actually, it's really not.  We know the FE isn't optimized for gaming yet (there are firmware updates coming from AMD for their AIB partners).  We know certain features are not enabled at the moment.  While I don't expect TBR to enhance performance significantly, it could help with the power draw at the expected speeds of the RX Vega cards (as opposed to the current power draw of Vega FE).  That can also be assisted by the firmware updates currently in the pipes.

 

How about this, we just agree to disagree until such time as the RX Vegas are actually in the hands of reviewers.  Deal?  I don't know what to expect from RX Vega, but what I do know is that I'm going to withhold judgment until the cards are actually available.  Speculation is all well and good, but speculation with a rush to judgment serves no one.  And your constant tirades against anyone who seemingly gives Vega the benefit of the doubt, is not helping matters.

[LAwLz]

4 minutes ago, Jito463 said:

Actually, it's really not.  We know the FE isn't optimized for gaming yet (there are firmware updates coming from AMD for their AIB partners).

I did not know that. Got any evidence to support the claim that Vega FE is not optimized for gaming?

Poor performance is not evidence for that by the way, since one explanation for that might just be that the card itself is not that good (which is not something I am claiming, but something I am saying could be an explanation).

Not using TBR is hardly "not optimized for gaming" by the way, since that should have next to no impact on performance. It would be far more accurate to say Vega Fe is not "optimized" for efficiency yet if all you are basing your statement on is TBR.

Again, Tile based rasterization is not a performance increasing feature. It's an efficiency feature (and I base that on what we can see in Nvidia, Imagination and ARM's implementation of it, which might not necessarily be exactly the same as AMD's, but it would be illogical to assume anything else*).

 

2 minutes ago, Jito463 said:

How about this, we just agree to disagree until such time as the RX Vegas are actually in the hands of reviewers.  Deal?

Not sure what we are disagreeing on.

I have never made any claims about the performance of Vega RX. The only thing I have said is that using Vega Fe to give a rough estimate of performance is a perfectly acceptable thing to do, and that you can not claim future updates will definitely improve performance.

 

12 minutes ago, Jito463 said:

 I don't know what to expect from RX Vega, but what I do know is that I'm going to withhold judgment until the cards are actually available.

That sounds like the good thing to do. I will withhold judgment too. I mean, we don't even know the price and that is a huge part of whether or not a product will be a good buy or not.  I think it's idiotic to make up your mind about a product before we got all of the crucial info.

 

13 minutes ago, Jito463 said:

Speculation is all well and good, but speculation with a rush to judgment serves no one.  And your constant tirades against anyone who seemingly gives Vega the benefit of the doubt, is not helping matters.

*There are two types of assumptions/speculations you can make.

1) An assumption based on what seems plausible based on other observations such as other products.

2) An assumption based on hopes and dreams.

 

Example of the two.

1) I don't expect Vega RX to perform better than Vega Fe because the workstation cards has historically not performed worse than the gaming cards for these types of tasks.

2) I expect Vega RX to perform better than Vega Fe. Fe is a "workstation card" and since it doesn't have gaming written on the box it is probably doesn't perform as well in games as Vega RX will.

[Jito463]

1 minute ago, LAwLz said:

Got any evidence to support the claim that Vega FE is not optimized for gaming?

AMD themselves have stated that Vega isn't their gaming card.  Whether you believe them or not is up to you, but that's direct from the manufacturer.

[LAwLz]

5 minutes ago, Jito463 said:

AMD themselves have stated that Vega isn't their gaming card.  Whether you believe them or not is up to you, but that's direct from the manufacturer.

"This isn't out gaming card" 

And

"Our gaming branded card will perform better in games than this card does" 

Are two completely different statements. 

 

I completely agree and trust AMD when they say FE is not their gaming card. What I don't agree with is that their gaming branded card is guaranteed to perform better in games. 

[Jito463]

1 minute ago, LAwLz said:

"This isn't out gaming card" 

And

"Our gaming branded card will perform better in games than this card does" 

Are two completely different statements. 

 

I completely agree and trust AMD when they say FE is not their gaming card. What I don't agree with is that their gaming branded card is guaranteed to perform better in games. 

Then, as I said, we'll have to agree to disagree until the actual cards are available.  You take the position that you don't believe them (as is your prerogative) , I take the position that I'm willing to wait for the final release with benchmarks.  We'll see what happens next month.

[LAwLz]

20 minutes ago, Jito463 said:

Then, as I said, we'll have to agree to disagree until the actual cards are available.  You take the position that you don't believe them (as is your prerogative) , I take the position that I'm willing to wait for the final release with benchmarks.  We'll see what happens next month.

Can you please post AMD's statement you claim I disagree with? 

Like I said before, I 100% believe and agree them when they say Vega FE is not their gaming card.

[Jito463]

1 minute ago, LAwLz said:

Can you please post AMD's statement you claim I disagree with? 

Like I said before, I 100% believe them when they say Vega FE is not their gaming card.

Seriously?  You posted the statement, now you want me to find the quote for you?  Sheesh.

 

http://wccftech.com/amd-gaming-rx-vega-will-be-faster-than-frontier-edition-showcase-confirmed-for-computex/

Quote

Raja Koduri – Chief Architect Radeon Technologies Group [Source]

Consumer RX will be much better optimized for all the top gaming titles and flavors of RX Vega will actually be faster than Frontier version!

 

[LAwLz]

6 hours ago, Jito463 said:

Seriously?  You posted the statement, now you want me to find the quote for you?  Sheesh.

I never posted that statement, nor have I said that I disagree with it.

I do find that statement hard to believe though, since there is no evidence whatsoever to support it. There is however quite a lot of evidence that suggests it will not be true.

[bomerr]

7 minutes ago, LAwLz said:

I never posted that statement, nor have I said that I disagree with it.

I do find that statement hard to believe though, since there is no evidence whatsoever to support it. There is however quite a lot of evidence that suggests it will not be true.

Ur wasting ur time. They are AMD Fanboys and don't want to accept the truth which is that AMD is far behind Nvidia. It's ironic because this RX Vega launch is repeating Fury X exactly. 

 

Fury X was hyped up, then results came out and it was on the level of a 980 without an OC and couldn't touch the 980 Ti, people said it'll get better and ya it still never beat the 980 Ti. Same situation with Vega, 1080 and 1080 Ti. 

[cj09beira]

4 hours ago, bomerr said:

Ur wasting ur time. They are AMD Fanboys and don't want to accept the truth which is that AMD is far behind Nvidia. It's ironic because this RX Vega launch is repeating Fury X exactly. 

 

Fury X was hyped up, then results came out and it was on the level of a 980 without an OC and couldn't touch the 980 Ti, people said it'll get better and ya it still never beat the 980 Ti. Same situation with Vega, 1080 and 1080 Ti. 

It's similar but not the same as Vega has features disabled/ not working right now plus the drivers, Fury X only had drivers as a way to improve perf and efficiency

[XenosTech]

2 minutes ago, cj09beira said:

It's similar but not the same as Vega has features disabled/ not working right now plus the drivers, Fury X only had drivers as a way to improve perf and efficiency

The amount of times I've been seeing you say this plus reviewers, it amazing that people just sweep it under the carpet instead of just waiting for RX vega to launch and then pass judgement then. Pretty sure most of these same people called Ryzen a failure before it even launched and look what happened.

[cj09beira]

1 minute ago, XenosTech said:

The amount of times I've been seeing you say this plus reviewers, it amazing that people just sweep it under the carpet instead of just waiting for RX vega to launch and then pass judgement then. Pretty sure most of these same people called Ryzen a failure before it even launched and look what happened.

I guess it goes against their narrative so they ignore it. 

Btw we need to pressure AMD to remove their Gpus bios checks ?

Imagine a world where you can change memory timings like you do for the ram, 

[XenosTech]

Just now, cj09beira said:

I guess it goes against their narrative so they ignore it. 

Btw we need to pressure AMD to remove their Gpus bios checks ?

Imagine a world where you can change memory timings like you do for the ram, 

I wouldn't want to do that but I'm sure that would make people like magetank really fucking happy

[cj09beira]

19 minutes ago, XenosTech said:

I wouldn't want to do that but I'm sure that would make people like magetank really fucking happy

After the mining craze disappears I will do some tests of the effect of timings on Gpus perf

[bomerr]

1 hour ago, XenosTech said:

The amount of times I've been seeing you say this plus reviewers, it amazing that people just sweep it under the carpet instead of just waiting for RX vega to launch and then pass judgement then. Pretty sure most of these same people called Ryzen a failure before it even launched and look what happened.

To be fair Ryzen is a failure for gaming. It has single threaded score of ~20% less than the 7700k and consistently preforms worse in gaming then the 7700k. The place where Ryzen excels is multithreaded applications as it's a far metter buy than X299. 

[Jito463]

12 hours ago, LAwLz said:

"This isn't out gaming card" 

And

"Our gaming branded card will perform better in games than this card does" 

Are two completely different statements. 

 

I completely agree and trust AMD when they say FE is not their gaming card. What I don't agree with is that their gaming branded card is guaranteed to perform better in games. 

11 hours ago, LAwLz said:

Can you please post AMD's statement you claim I disagree with? 

Like I said before, I 100% believe and agree them when they say Vega FE is not their gaming card.

5 hours ago, LAwLz said:

I never posted that statement, nor have I said that I disagree with it.

I do find that statement hard to believe though, since there is no evidence whatsoever to support it. There is however quite a lot of evidence that suggests it will not be true.

5 hours ago, bomerr said:

Ur wasting ur time. They are AMD Fanboys and don't want to accept the truth which is that AMD is far behind Nvidia. It's ironic because this RX Vega launch is repeating Fury X exactly. 

So, saying we should reserve judgment until the RX cards arrive is being a fanboy?  If anyone sounds like a fanboy, it's you.

[Jito463]

1 minute ago, bomerr said:

To be fair Ryzen is a failure for gaming. It has single threaded score of ~20% less than the 7700k and consistently preforms worse in gaming then the 7700k. The place where Ryzen excels is multithreaded applications as it's a far metter buy than X299. 

You're operating off of old information.  Many games have been updated to provide significantly closer performance numbers, and there's also the Nvidia drivers that don't run as well on Ryzen as they do on Intel.  There's not that large of a gap.

[bomerr]

13 minutes ago, Jito463 said:

You're operating off of old information.  Many games have been updated to provide significantly closer performance numbers, and there's also the Nvidia drivers that don't run as well on Ryzen as they do on Intel.  There's not that large of a gap.

I don't believe you and the reason why is A) benchmarks, which you consider old okay... B) I just overlocked my CPU from 4.2 to 4.6 GHz and even only hitting ~70% cpu load, my FPS increased so games do benefit from higher single threaded performance right now and C) Kaby Lake performs better in gaming than Skylake-X because the architecture better maximizes single score performance.