MSI GeForce RTX 2060 Gaming Z 6 GB

[W1zzard]

MSI's GeForce RTX 2060 Gaming Z is the best RTX 2060 custom-design we've reviewed so far. It comes with idle-fan-stop and a large triple-slot cooler that runs cooler than the Founders Edition. Noise levels are excellent, too; it's the quietest RTX 2060 card to date.

Show full review

 

[Noztra]

I think your RX Vega benchmark are broken?

Other sites seems to have a fair bit higher Vega benchmarks than yours? In your benchmarks the Vega 64 in most games gets smoked by the 1070ti. Other sites the have the Vega 56 going toe to toe with the 1070ti and some even beating it, and now the 1070ti completely destroys both Vega 56/64?

And if you compare a AIB, then it should be towards Vega AIB's as well? You only have "stock" Vega's which thermal throttles a lot. Sapphire's Vega 56 is for instance just a fast as the LC Vega 64. I know that you have stock RTX2060 in your benchmarks as well, but would be nice to see AIB Vega's as well.

Not taking anything from all the work put into this review and other reviews, but the Vega benchmarks seems flawed and not correct compared to others.

Thank you.

/Noz

Edit: Even compared to your own stock Vega 64 review, the numbers seems off?

 

[Assimilator]

I think your RX Vega benchmark are broken?

Other sites seems to have a fair bit higher Vega benchmarks than yours? In your benchmarks the Vega 64 in most games gets smoked by the 1070ti. Other sites the have the Vega 56 going toe to toe with the 1070ti and some even beating it, and now the 1070ti completely destroys both Vega 56/64?

And if you compare a AIB, then it should be towards Vega AIB's as well? You only have "stock" Vega's which thermal throttles a lot. Sapphire's Vega 56 is for instance just a fast as the LC Vega 64. I know that you have stock RTX2060 in your benchmarks as well, but would be nice to see AIB Vega's as well.

Not taking anything from all the work put into this review and other reviews, but the Vega benchmarks seems flawed and not correct compared to others.

Thank you.

/Noz

I wish people would read the "Test Setup" page, in particular the version of AMD Catalyst driver used. The reason for using such an old version is also given below the table on that page.

tl;dr:

• W1zz last reviewed an AMD GPU when Catalyst 18.8.2 was available
• he only changes driver versions when he tests a new card (or when a specific driver version is required for a card, e.g. RTX)
• the reason for this is that he has to re-benchmark ALL AMD/NVIDIA cards on that new driver version for the comparisons to be fair
• he has a real job and other time constraints, which means re-benching is out of the question every time AMD or NVIDIA release a new driver
• there is a lot of flux in the video card market (many new products being released) which means that it's pointless to re-bench until it has settled down, as new drivers are coming thick and fast
• he will probably re-bench all cards on their manufacturers' latest driver versions once the Radeon VII is launched (since that is likely to be the last big launch for a while)
• he is not going to go back and update all his reviews once the re-bench is complete, again because of time constraints

Yeah, it's an imperfect process, but we live in an imperfect world. That's why you get your information from more than one source.

 

[Tsukiyomi91]

so, an extra $30 something premium over FE for a marginal gain at 1080p but starts gaining more ground on 1440p... this TU106 core really loves high resolution... so, we crown it as the new 1440p GPU king for under $400? XD

 

[ZeroFM]

Still 2060FE is best choice

 

[Noztra]

I wish people would read the "Test Setup" page, in particular the version of AMD Catalyst driver used. The reason for using such an old version is also given below the table on that page.

tl;dr:

• W1zz last reviewed an AMD GPU when Catalyst 18.8.2 was available
• he only changes driver versions when he tests a new card (or when a specific driver version is required for a card, e.g. RTX)
• the reason for this is that he has to re-benchmark ALL AMD/NVIDIA cards on that new driver version for the comparisons to be fair
• he has a real job and other time constraints, which means re-benching is out of the question every time AMD or NVIDIA release a new driver
• there is a lot of flux in the video card market (many new products being released) which means that it's pointless to re-bench until it has settled down, as new drivers are coming thick and fast
• he will probably re-bench all cards on their manufacturers' latest driver versions once the Radeon VII is launched (since that is likely to be the last big launch for a while)
• he is not going to go back and update all his reviews once the re-bench is complete, again because of time constraints

Yeah, it's an imperfect process, but we live in an imperfect world. That's why you get your information from more than one source.

That just highlight my point even more?

If you wanna compare to competing products, you have to make sure they are competing on equal terms otherwise the results are flawed?

You have to compare apples to apples, otherwise you are giving product X ( In this case AMD and old NVIDIA cards) a severe disadvantage.

Its like comparing a BMW vs Mercedes in a 400m drag race and the Mercedes is running on flat tires. Conclusion the BMW wins, which is technical correct, but not really objective.

And popular site like TPU which has many readers, must have an interest in their reviews are correct and objective?

But tbf I am not saying that Wizzard should redo every old graphics card everytime he does a test. But it could be an idea to redo some of them once every 6 months or something to get more accurate results. Because atm the conclusion (Gaming performance exceeds Vega 64 and GTX 1080 ) is pretty useless.

 

[W1zzard]

the reason for this is that he has to re-benchmark ALL AMD/NVIDIA cards on that new driver version for the comparisons to be fair

Just finished a full rebench with new games and new drivers, will be used starting with the MSI 2080 Ti Lightning Z review.

You only have "stock" Vega's which thermal throttles a lot.

Maybe that can explain it? I heat up ALL cards before starting benchmark result recording, to reflect actual long gameplay, not a 30 second benchmark session with a cold card that will boost sky high. This affects both AMD and NVIDIA btw

In your benchmarks the Vega 64 in most games gets smoked by the 1070ti

Looking at the average they are almost exactly identical in performance

I'm sure you'll be happy to hear that in the next rebench, due to different games, Vega64 is now faster by a few % than GTX 2060 FE

 

[jabbadap]

Still 2060FE is best choice

Quite so, yes. Only thing missing on it is FAN-stop idle and maybe it TDP limits are very strict. But overall FE is very solid card, with the best video outputs, very silent cooling and yeah it's not even look that bad(not that really matters).

Was this the first card, which could beat FE in noise levels?

 

[Noztra]

Just finished a full rebench with new games and new drivers, will be used starting with the MSI 2080 Ti Lightning Z review.

Cool, thank you.

Maybe that can explain it? I heat up ALL cards before starting benchmark result recording, to reflect actual long gameplay, not a 30 second benchmark session with a cold card that will boost sky high. This affects both AMD and NVIDIA btw

I agree, but then maybe still should be some AIB Vega's which don't have the problem since you are comparing them to NVIDIA AIB's.

Looking at the average they are almost exactly identical in performance

I'm sure you'll be happy to hear that in the next rebench, due to different games, Vega64 is now faster by a few % than GTX 2060 FE

I am not happy because its AMD, because I would have wrote the same if it was the other way around.

But again thank you, for your reply.

 

[gamerman]

excellent gpu! it just show clear hoe lausy junks all polaris and vega gpus are... phyi!

i hate one thing alot nvidia rtx 2000 series,and that it not support windows 7!

come on! ppl must send alot messages and notes nvidia and its AIB partners.... we want rtx 2000 series support window 7! l ts do it...

im sure it can do with drivers.


lets go!

 

[0x4452]

i hate one thing alot nvidia rtx 2000 series,and that it not support windows 7!

They certainly do support Win7. DX12 does not support Win7. Thankfully ray tracing will be available through Vulkan.

I am personally staying on Win7 for the foreseeable future

 

[rtwjunkie]

I agree, but then maybe still should be some AIB Vega's which don't have the problem since you are comparing them to NVIDIA AIB's.

Whether it is AMD or Nvidia, when W1z tests older cards against newer releases, almost always the older cards are not AIB. They are reference. Since he retests all cards, it is unreasonable to expect more than the base models for comparison. That could end up being 100 cards.

If you are using multiple review sites, and go back and look at his review of the AIB model you wish to see compared (back when it was released) and you can get a good idea where that would sit in the current comparisons.

 

[swirl09]

so, an extra $30 something premium over FE for a marginal gain at 1080p but starts gaining more ground on 1440p...

Forget the marginal gains, the $30 for that cooler is worth it. At least, Id consider it money well spent if I was in the market for one.

 

[bug]

Forget the marginal gains, the $30 for that cooler is worth it. At least, Id consider it money well spent if I was in the market for one.

I'm not so sure about that. I'm looking to get a 2060, I wouldn't pay $30 for a fan that's basically as loud as the reference cooler where it matters.
Still, this has one more DP and dropped the ancient DVI which is smart imho. But they also dropped the USB-C and one phase for the VRM (doesn't seem to affect performance at all, might affect longevity). I honestly don't know what to pick. Currently I'm thinking MSI Ventus, Gigabyte's entry level or FE.

 

[RichF]

If a company releases a card with ineffective reference cooling then it deserves what it gets. It's called reference for a reason.

However, if 3rd-party versions offer much better performance due to something as stupid as throttling due to the use of an ineffective cooler in the reference model, it would be a good idea to note that in a footnote.

 

[RH92]

so, an extra $30 something premium over FE for a marginal gain at 1080p but starts gaining more ground on 1440p

You don't pay those extra 30 bucks for pure perf , i mean sure you do but that's not the main reason , the main reason is better thermals and noise .... wich this card delivers .

Don't get me wrong that FE cooler is impressive non the less being this close to such an excellent custom cooler .

 

[Tsukiyomi91]

I know. also, I didn't say it was a bad card. It all depends really. >< aftermarket cooler is always a good thing if you want to push the core a little more.

 

[JRMBelgium]

Another fake chart. Let's compare the RTX 2060 witht the latest driver, to AMD hardware tested on 5-month old drivers. And then, let's conclude that the latest RTX 2060 is faster then Vega 64.
But who cares about correct figures….

And before the "it doesn't change much" comments follow. I just recently got almost 30% performance increase on Forza 7 by upgrading to driver 19.1.1. Don't beleave me, download one of the 2018 drivers, run the game, download 19.1.1, run the game again. And I know the game is not included on TPU. But Wolfenstein, BF1 and Strange brigade are included and all of them had performance improvements in the last months… As a gamer, you know this. As a reviewer...who cares right?

 

[Fluffmeister]

Frankly it is a sad state of affairs to have AMD's Vega 64 trading blows with a 106 tier, xx60 card. A W1zz noted Vega 64 will regain the lead by a whooping couple of percent with a rebench and change of games.

If you don't like it start your own site and get benching, I wish i could recall his name, but one dude wanted to exclude all the games Nvidia does better in, get him to sign up too... you know, fight the power!

 

[ZoneDymo]

Why is the power consumption of this card at idle 9 watt and a GTX1080 (previous gen high end model) 6 watt.....

feels like we are going backwards.

 

[xkm1948]

If you don't like it start your own site and get benching, I wish i could recall his name, but one dude wanted to exclude all the games Nvidia does better in, get him to sign up too... you know, fight the power!

Man that is just sad. Gotta be a real loser IRL to focus so much on justfying a gaming GPU's existence considering neither AMD nor Nvidia are paying them to do so. Just sad.

 

[Tsukiyomi91]

Independent reviews is no easy task, considering AMD, Intel, Nvidia & their AIB vendor partners are not very keen on giving away their samples to those who may keep the product or modify it to the point of breaking their products... on top of ensuring you have the highest end hardware of a bench rig to prevent ANY bottleneck or variations when doing a benchmark to ensure consistency. Also, you must set a standard in your testing methodology or no one will look your tabulated data & calls it "BS".

 

[bug]

Frankly it is a sad state of affairs to have AMD's Vega 64 trading blows with a 106 tier, xx60 card. A W1zz noted Vega 64 will regain the lead by a whooping couple of percent with a rebench and change of games.

If you don't like it start your own site and get benching, I wish i could recall his name, but one dude wanted to exclude all the games Nvidia does better in, get him to sign up too... you know, fight the power!

Tbh, 2060 is a strange beast. It's priced towards the high end, it performs like the high end. And yet, with no less than 3 other cards in Nvidia's lineup performing (significantly) better, we can't really call it high end.
It's mid range at the end of the day, just not at a price point we're comfortable with for the mid range

 

[Tsukiyomi91]

Indeed. Strange to the point that I want it even more. For a FE card worth $350 performing as close to the 1070Ti at 1440p wasn't interesting enough, I'm a little lost at why people are throwing a hissy fit at a new card that just charted the previously uncharted waters, for that kind of money & still pulls an impressive bench, both stock & OCed.

@ZoneDymo reason why Turing core has slightly higher draw is the fact it has 2 newer additional features baked into the core; RT Cores & Tensor Cores. Pascal do not have those features hence the "odd" comparison. Also, Pascal can't even push 1 gigarays/sec. Also, I doubt 9W idle draw will affect one's total system draw or one's monthly electricity bills.

 

[bug]

Indeed. Strange to the point that I want it even more. For a FE card worth $350 performing as close to the 1070Ti at 1440p wasn't interesting enough, I'm a little lost at why people are throwing a hissy fit at a new card that just charted the previously uncharted waters, for that kind of money & still pulls an impressive bench, both stock & OCed.

@ZoneDymo reason why Turing core has slightly higher draw is the fact it has 2 newer additional features baked into the core; RT Cores & Tensor Cores. Pascal do not have those features hence the "odd" comparison. Also, Pascal can't even push 1 gigarays/sec. Also, I doubt 9W idle draw will affect one's total system draw or one's monthly electricity bills.

Well, if AMD gives you no reason to praise their video cards, one can always talk thrash about Nvidia's. It's more convenient than admitting AMD is lagging in perf/W, have failed to implement TBR so far and now they have also taken a step back in RTRT and AI assisted rendering. As much as I root for Navi, it's hard to imagine AMD closing all those gaps in just one iteration.