AMD Radeon VII Then vs. Now Gaming Performance

One thing that changed a lot from launch to this date is that many more programs have received optimized code for tensor cores.
At launch i was prone to get a Radeon VII against a 2080.
Today i would never dream of going Red, because some things i use a lot run much faster on RTX cards.
An the list of things that tensor cores can speed up will only increase.
 
Great Review! Love the concentrated effort with one video card, straight to the point and reminds me of the quality level of Wavy Dave back in the day at Beyond 3d.

Also the inbuilt benchmark used at times does allow a comparison to one own's machine to see if the upgrade would be worth it. So that is what I did.
While not a apple to apple comparison, it is a apple to apple game benchmark test, very refreshing in that one can participate in an evaluation. Also the review doing real work of finding any other issues by actually playing the game for the real scoop is pure gold.
  • AMD 2700 all cores at 4.1ghz, ASUS C6H, 32gb of DDR 3200 at DDR 3200, Win 10 64bit
  • Radeon Vega 64 LC, mem OC to 1025, Power level +25, 19.5.2 drivers
  • Shadow of the Tomb Raider results:
    • All Max Settings, 1440p with HDR, SMAAT2X (same as review)
      • Inbuilt benchmark - 66 FPS - Vega Vii stock is 12% faster
    • My HDR Settings; All max except no motion blur, Normal Depth of field vice High
      • 76 FPS, which has performance the same but with less IQ than the Vega Vii review (non OC)
        • TAA has slightly better motion AA but blurrier textures, SMAAT2X to me is higher quality AA but with a performance cost
If one has a Vega 64, going to a Vega Vii is sort of a hard sell for just games. Other aspects may make it very much a great upgrade - can't wait to see how that plays out in the coming review. Plus OCing or potential the card has I hope to see as well.
 
Great read!!! Thanks for the dive. The driver performance changes are interesting. I would think that performance increases would be more equal across different resolutions unless the driver itself was being worked on(it's use of threading, etc...), compared to specific optimizations.

I wish someone with one of these could do some Destiny 2. At 3840x1600, it can be a challenging game to drive with everything cranked. My 1080Ti mostly cuts the mustard, but I am curious about the RVII since nVidia went and broke their VRR drivers.
 
A couple of things.

One, this is major. You NEED a drop-down index for the review on every page so that we can easily navigate through the article. Page numbers are nice, but meaningless because I won't know what's where.

Secondly, make an effort. What do I mean? In the DX:MD section you state the following "However, once we get to 4K, performance drops off and the game is unplayable. " First of all, 40 fps is far from unplayable. Tone down the language & be less extreme. It might not be optimal, but it is above the minimum acceptable standard and going into smooth-ish territory. More importantly, by going with the ultra preset you are severely gimping performance due to it choosing CHS Ultra, which has almost no visual impact at all but which murders framerates. If you disable this, the card will give you >50 fps at least, making it much more playable and invalidating your previous assertion that the game is "unplayable" at 4K on a VII.

Same thing, with KC:D. Where's the talk of r_batch type change? You are short changing your readers by doing lazy testing, like everyone else, just through the presets.

You already test few games, at least get to know the games you do test so you can better serve your audience. Tbh, right now there's very little reason to bother reading just another tech review site, when it does very little extra or different.

Also, no overclocking results? Yikes!

Kudos for the HDR testing though, that's something.

You have a chance to build something here, but it's up to you whether you want to do it. I hope you will.
 
Accedie said:
A couple of things.

One, this is major. You NEED a drop-down index for the review on every page so that we can easily navigate through the article. Page numbers are nice, but meaningless because I won't know what's where.

Secondly, make an effort. What do I mean? In the DX:MD section you state the following "However, once we get to 4K, performance drops off and the game is unplayable. " First of all, 40 fps is far from unplayable. Tone down the language & be less extreme. It might not be optimal, but it is above the minimum acceptable standard and going into smooth-ish territory. More importantly, by going with the ultra preset you are severely gimping performance due to it choosing CHS Ultra, which has almost no visual impact at all but which murders framerates. If you disable this, the card will give you >50 fps at least, making it much more playable and invalidating your previous assertion that the game is "unplayable" at 4K on a VII.

Same thing, with KC:D. Where's the talk of r_batch type change? You are short changing your readers by doing lazy testing, like everyone else, just through the presets.

You already test few games, at least get to know the games you do test so you can better serve your audience. Tbh, right now there's very little reason to bother reading just another tech review site, when it does very little extra or different.

Also, no overclocking results? Yikes!

Kudos for the HDR testing though, that's something.

You have a chance to build something here, but it's up to you whether you want to do it. I hope you will.
Click to expand...
The review was about Driver performance differences from the first to the current. That was the focus. In other words make sure you understand the focus of the review.

Radeon Vii has some improvements over time due to drivers but nothing earth shattering. Not an article in how to adjust your game settings to hit X amount of FPS, that would be more particular game related articles that Brent has done over the years. As for 40 FPS as unplayable, for some that is utterly true for FPS and MP games. For me FreeSync above 40 FPS is fine but prefer over 60 FPS when I can get it.

If one is thinking about buying a Radeon Vii presently, it is nice to know how it actually performs maxed out at the different resolutions which Brent clearly shows with all firmware and driver improvements. This is with the new games, many with DX 12 enabled. HDR coverage I agree was sweet.
 
alxlwson said:
Great read!!! Thanks for the dive. The driver performance changes are interesting. I would think that performance increases would be more equal across different resolutions unless the driver itself was being worked on(it's use of threading, etc...), compared to specific optimizations.

I wish someone with one of these could do some Destiny 2. At 3840x1600, it can be a challenging game to drive with everything cranked. My 1080Ti mostly cuts the mustard, but I am curious about the RVII since nVidia went and broke their VRR drivers.
Click to expand...

I can tell you that you can't always get 60FPS with the game maxed out on an overclocked RTX 2080 Ti at 3840x2160. I'm an avid Destiny 2 player so I know how it performs on 1080Ti's and the RTX 2080 Ti. I don't know about playing it at 3840x1600. That's an odd resolution and we can't possibly cater to every weird resolution out there.
 
Accedie said:
One, this is major. You NEED a drop-down index for the review on every page so that we can easily navigate through the article. Page numbers are nice, but meaningless because I won't know what's where.
Click to expand...

Agree - I've been trying to figure this one out, but it's not easily supported by this theme/Wordpress in general. I hope I'll get it sorted soon enough.

Accedie said:
Secondly, make an effort. What do I mean? In the DX:MD section you state the following "However, once we get to 4K, performance drops off and the game is unplayable. " First of all, 40 fps is far from unplayable. Tone down the language & be less extreme. It might not be optimal, but it is above the minimum acceptable standard and going into smooth-ish territory. More importantly, by going with the ultra preset you are severely gimping performance due to it choosing CHS Ultra, which has almost no visual impact at all but which murders framerates. If you disable this, the card will give you >50 fps at least, making it much more playable and invalidating your previous assertion that the game is "unplayable" at 4K on a VII.
Click to expand...

@Brent_Justice can probably provide a bit more light on this, but as @noko mentioned, the main goal of the article is to compare performance changes since the launch of the Radeon VII, so we elected to do so in a fairly standardized manner. As we're just getting started, we have not yet produced any of our game play reviews, and those are where we will focus on the best settings for a given game and differentiate between the presets, what they bring to the table and even image quality comparison sliders.

From a "playable" perspective, this is something that's always subjective - some games may feel just fine at 25-30FPS while others may feel horrible at 60FPS. It's something that people have attempted to measure over the years (see: FCAT for frame time analysis), but honestly, it comes down to a gamer playing the game and feeling whether the game is appropriately responsive to his/her inputs.

Hopefully you'll see our differentiation over time - we've spent a while planning out the types of content you'll see on the GPU side of the house and I hope you like it as we get rolling!
 
THUMPer said:
Hot ****, what a great review! So Glad to see a 2700x being used.
Click to expand...

Thank you, I thought it was important to utilize the latest from AMD, they are great on price and value, and with the next gen in sight it is important to have those comparisons from the old to the new.
 
noko said:
Great Review! Love the concentrated effort with one video card, straight to the point and reminds me of the quality level of Wavy Dave back in the day at Beyond 3d.

Also the inbuilt benchmark used at times does allow a comparison to one own's machine to see if the upgrade would be worth it. So that is what I did.
While not a apple to apple comparison, it is a apple to apple game benchmark test, very refreshing in that one can participate in an evaluation. Also the review doing real work of finding any other issues by actually playing the game for the real scoop is pure gold.
  • AMD 2700 all cores at 4.1ghz, ASUS C6H, 32gb of DDR 3200 at DDR 3200, Win 10 64bit
  • Radeon Vega 64 LC, mem OC to 1025, Power level +25, 19.5.2 drivers
  • Shadow of the Tomb Raider results:
    • All Max Settings, 1440p with HDR, SMAAT2X (same as review)
      • Inbuilt benchmark - 66 FPS - Vega Vii stock is 12% faster
    • My HDR Settings; All max except no motion blur, Normal Depth of field vice High
      • 76 FPS, which has performance the same but with less IQ than the Vega Vii review (non OC)
        • TAA has slightly better motion AA but blurrier textures, SMAAT2X to me is higher quality AA but with a performance cost
If one has a Vega 64, going to a Vega Vii is sort of a hard sell for just games. Other aspects may make it very much a great upgrade - can't wait to see how that plays out in the coming review. Plus OCing or potential the card has I hope to see as well.
Click to expand...

Thank you for recognizing this, utilizing the built-in benchmarks that games have means everyone can get involved and compare results with our own, I think it's brilliant! We want everyone to be involved in our reviews with us, I want to see everyone else's results.
 
Accedie said:
A couple of things.

One, this is major. You NEED a drop-down index for the review on every page so that we can easily navigate through the article. Page numbers are nice, but meaningless because I won't know what's where.

Secondly, make an effort. What do I mean? In the DX:MD section you state the following "However, once we get to 4K, performance drops off and the game is unplayable. " First of all, 40 fps is far from unplayable. Tone down the language & be less extreme. It might not be optimal, but it is above the minimum acceptable standard and going into smooth-ish territory. More importantly, by going with the ultra preset you are severely gimping performance due to it choosing CHS Ultra, which has almost no visual impact at all but which murders framerates. If you disable this, the card will give you >50 fps at least, making it much more playable and invalidating your previous assertion that the game is "unplayable" at 4K on a VII.

Same thing, with KC:D. Where's the talk of r_batch type change? You are short changing your readers by doing lazy testing, like everyone else, just through the presets.

You already test few games, at least get to know the games you do test so you can better serve your audience. Tbh, right now there's very little reason to bother reading just another tech review site, when it does very little extra or different.

Also, no overclocking results? Yikes!

Kudos for the HDR testing though, that's something.

You have a chance to build something here, but it's up to you whether you want to do it. I hope you will.
Click to expand...

Thank you for the reply.

On the topic of "what is playable" while the average may show 40 you have to consider the minimums are a lot less, a game needs to at least maintain a certain minimum framerate, and play smoothly. So when we state something isn't playable, it's not just the average FPS number we are going off of, but actually playing the game and feeling it out and our experiences of playing it help determine if its playable or not.

I agree that lowering settings will make it more playable. However, this was not the scope or goal of the article. This was a video card review review, focused on the hardware. In our gameplay evaluation reviews, focused on the game, is where we will dive into "what it takes" to make things playable by lowering settings at each resolution.

Again, we do go beyond the presets, in many games the highest presets are not actually the highest game settings. We do manually configure the settings to attain the highest image quality in games. In our game reviews, we will explore custom game settings to find what is playable. Two different reviews with two different goals.

We actually test quite a bit of games, more so than I usually tested previously. It does take time.

This was the first part of a multi-part review. This review was also focused on performance over time, which doesn't include overclocking. It wasn't a traditional "review" of the video card as much as it was a review of how performance has or has not changed over time. We do plan to do more with the Radeon VII, but our articles are very focused on certain topics/goals. It's important to understand that and so we do explain it on the test setup page. I will most likely do an overclocking focused article in the future.

We have many things planned out right now, it cannot all come at once we need the time to test and write them. We are taking the approach of articles focused on specific topics, this way everything gets the most focus and testing and evaluation we can give it. Bare with us, stay tuned, and you may see some things that you like.
 
Love the article.....it really covers some facts i was wandering about. (not sure why i didnt see it till today)
 
You must log in or register to reply here.
Become a Patron!
Back
Top