[Hardware Unboxed] Ryzen 2400G/2200G Vega Graphics Performance

[dogen1]

The Xbox One has 68.4 GB/s of memory bandwidth, as it uses DDR3. Even if we assume dual channel DDR4-2400, Raven Ridge has more than 1/2 the Xbox's memory bandwidth. If we assume dual channel DDR4-3200, it has 3/4 the memory bandwidth. The PS4 has a lot more of course, which is probably what you meant.

I'm guessing that Vega 11, even with the lower memory bandwidth is ahead of the Xbox One on average. It's pretty close to the GT 1030, which must be faster than a downclocked DDR3 R7 260.

You're forgetting the high bandwidth (somewhere around 200GB/s peak) ESRAM on the xb1 which is extensively utilized by games.

 

[frozentundra123456]

Based on one game?

This is post is definition of bias, and Anti-AMD approach.

Well, if you dont like the results posted, follow the provided link and read the entire test. Results for several more games and synthetics are provided.

 

[frozentundra123456]

An i3 8100 and a Geforce 1050 would be $270 here in the US. The 2400G is $180, and the difference between 8GB 2400 DDR4 and 3200 is only $20 or so. So while the 1050 definitely performs better, it's also definitely more expensive. If you have an eye for adding a better graphics card further down the road, I'd rather have the unlocked 2400G with SMT to pair with it rather than the locked 8100 with no SMT.

If you are looking to add even a moderately powerful dgpu, either the Ryzen 1600 or i5 8400 (when cheaper mb are available) is then a better buy than the 2400. Only about 30.00 more and much better cpu performance or gaming performance with dgpu.

 

[VirtualLarry]

I'm talking about DDR4 vs not DDR4.

I'm not really sure what the point that you're trying to make is, since no current modern (desktop) platform or motherboards, use DDR3 or anything lower than DDR4. (Except maybe some Atom ITX boards that might use DDR3L SO-DIMMs.)

 

[24601]

I'm not really sure what the point that you're trying to make is, since no current modern (desktop) platform or motherboards, use DDR3 or anything lower than DDR4. (Except maybe some Atom ITX boards that might use DDR3L SO-DIMMs.)

No need to play dumb, we've talked about this exact issue in every single thread that you have created for the last 2 years.

 

[VirtualLarry]

Oh, that used, beat-up, last-gen systems use last-gen DRAM technology, and thus cheaper to build with? Is that your point? I don't like having to guess here.

I believe your original point, was simply that the 2200G / 2400G are too expensive for entry-level builds, simply because of the cost of DDR4. I don't buy that premise.

But that cost of RAM is the same between AMD and Intel rigs, for the most part, except for vendors possibly gouging prices on "Ryzen Compatible" RAM which seems to have a premium that the cheaper "XMP Intel compatible" DDR4-3000 and up don't have.

Edit: I was simply pointing out, that the "DDR4 or not-DDR4" question, wasn't even a valid one, for people that prefer to "build new" - something that's perfectly acceptable, to only want "new" hardware, if you purchase for longevity, rather than saving every last $$$ and aren't afraid of pot-luck hardware lifespans.

Edit: To wit, I sold a friend of mine a "refurb gaming PC" that I built, the first one I built, based on an ebay-purchased OEM refurb office tower PC, with a quad-core Intel CPU. I installed 16GB DDR3-1600, a new SSD, a new 1TB HDD, and a new GTX1050.

Well, he's having problems with it, "kernel error" with the video drivers, when he watching several YouTube vids.

Fresh OS install, no issues that I know of with all of the new parts (although, I will have him run a memory test, if I didn't already when I built it).

But I'm really regretting going the "used, refurb, it's cheaper!" route, for his "new" gaming rig, if it's giving him trouble, that could indicate a bum PSU or mobo, or even CPU.

Edit: I guess my point with that anecdote was, building using refurb parts / chassis / mobo / CPU, can be "cheaper", in the absolute dollar sense, but it may not actually be "better".

I mean, performance of it is fine, it's just that I don't think that it should be crashing while watching YouTube.

It could be a browser / video driver issue.The curious thing is, it happens on both his "ancient, falling-apart" rig, with an Intel Core2Quad CPU, Windows 7 64-bit, and a 2GB Fermi GTX460, and the same thing (he claims), is happening with his Ivy Quad-core rig, with a new GTX1050 2GB, also with Win7 64-bit.

I'll have to dig more info out of him, maybe it's a software issue after all, or some interaction with his Browser.

Might ask him to switch browsers for a week or two and see if the crashes / kernel driver re-starts go away.

Or maybe the refurb OEM box needs a new PSU. If it's ATX, I'll just go over and throw a brand-new one in for him.

Edit: If your point, was simply that for budget gaming builds, building NEW with one of the new Ryzen APUs, was more expensive due to DDR4 prices, than buying and building a gaming PC out of refurb parts, I would partially agree that it could be more expensive, but given my experiences, of apparently losing the "refurb silicon lottery" with my friend's rig, sometimes newer and brand-NEW is better, even if it isn't actually cheaper. (Since everyone equates APU with "cheaper, like a console".)

Really, it kind of comes down to the individual refurb quality, which can be hard for an ordinary online purchaser that can't directly inspect the PC before purchase and run it, to determine whether it's worth purchasing.

Edit: I do agree that DDR4 is WAY overpriced, though. That was one of my motivations to build my (first) FX-8320E rig, on a Gigabyte 78LMT-USB3 R2 mobo, because I already had a slightly-used DDR3-1600 kit. I also bought the Team Group L5 Lite 3D SSD, the 480GB one that's $100, both then, and currently now at Newegg, probably will expire tomorrow. Topped off with a GT1030.

Yeah, I want a 2200G (2400G is hard to swallow, price-wise) APU too, but the DDR4 prices ARE kind of a killer. But you can always go with an 8GB kit if you really HAVE TO. Which I probably will.

Already have a Gigabyte Gaming 3 B350 mATX board for my 2200G APU.

 

[Red Hawk]

If you are looking to add even a moderately powerful dgpu, either the Ryzen 1600 or i5 8400 (when cheaper mb are available) is then a better buy than the 2400. Only about 30.00 more and much better cpu performance or gaming performance with dgpu.

The 2400G isn't going to bottleneck a 1060/580 like a Pentium G4560 was my point. The Ryzen 1600/i5 8400 are better CPUs, sure, but the 2400G is enough. And you're looking at spending well over $100 more to get a Ryzen 5 1600/i5 8400 with a GPU that's as good or better than the Vega 11. The whole point of the 2400G is the low price of entry.

 

[Glo.]

Well, if you dont like the results posted, follow the provided link and read the entire test. Results for several more games and synthetics are provided.

Did you even checked reviews from other sites? If yes, you would agree with me, and not post this.

 

[Muhammed]

The 1030 and Vega 11 maybe close to each other in several games, but the 1030 absolutely trounces the Vega 11 in several other titles!

https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,11
https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,15

 

[PingSpike]

Meet the walking dead.
<snip>

No idea why people are still after the i7 7700K. Its grossly overpriced when selling at MSRP now that coffeelake is on the scene and even has good availability. I wouldn't even buy it if I had already had a z200 series board unless it had a steep discount personally.

But then again I think the AMD dozer 8 cores were still hanging out in that list until Ryzen showed up and those things were horrible so I guess craziness abounds.

 

[maddie]

Your definition of "dead" is a little bit off:

https://pclab.pl/art76962-11.html

Here is another popular game: https://techreport.com/review/33235/amd-ryzen-3-2200g-and-ryzen-5-2400g-processors-reviewed/6

If somebody would buy new the combination of Pentium G4560 and dGPU is still better.

You're BACK. Some very selective results as was the norm. I guess we're in for "interesting" posts again.

Have to be diplomatic, after all.

 

[Red Hawk]

The 1030 and Vega 11 maybe close to each other in several games, but the 1030 absolutely trounces the Vega 11 in several other titles!

https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,11
https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,15

I'm supremely suspicious of the Starcraft 2 results there. Not only are the Vega graphics way behind both the GT 1030 and the RX 550, but they're behind the Intel HD Graphics 630, which is using slower memory! Judging by ALL other tests, that just shouldn't happen. And it's not simply AMD being terrible at the game, since the 550 does fine. That's indicative of a testing or driver issue, if I ever saw one.

 

[24601]

I'm supremely suspicious of the Starcraft 2 results there. Not only are the Vega graphics way behind both the GT 1030 and the RX 550, but they're behind the Intel HD Graphics 630, which is using slower memory! Judging by ALL other tests, that just shouldn't happen. And it's not simply AMD being terrible at the game, since the 550 does fine. That's indicative of a testing or driver issue, if I ever saw one.

It's obviously a CPU bottleneck.

 

[Qwertilot]

Really? The i3-8100 is the fastest intel CPU there & shouldn't the 22/400g be at least close to the G4560 as pure CPU's?

Seems more like some random oddness.

 

[24601]

Really? The i3-8100 is the fastest intel CPU there & shouldn't the 22/400g be at least close to the G4560 as pure CPU's?

Seems more like some random oddness.

I'm talking about the AMD APU vs intel CPU + 550

UHD 630 is obviously GPU bottlenecked.

2400/2200g is just worse ryzen with less cores and less clocks (as well as getting reamed by memory bandwidth due to having to share with the iGPU)

 

[Red Hawk]

I'm talking about the AMD APU vs intel CPU + 550

UHD 630 is obviously GPU bottlenecked.

2400/2200g is just worse ryzen with less cores and less clocks (as well as getting reamed by memory bandwidth due to having to share with the iGPU)

So a 3.9 GHz max boost, quad core/eight thread Ryzen can't hit 30 FPS in Starcraft, and craters to the teens for minimums, but a 3.5 GHz dual core/four thread Kaby Lake CPU has no problem passing 40 FPS? I'm skeptical, to say the least. And if memory bandwidth was the issue, then I'd expect the UHD 630 to tank even worse, since it's running with slower memory than they tested the 2400G and 2200G with.

 

[24601]

So a 3.9 GHz max boost, quad core/eight thread Ryzen can't hit 30 FPS in Starcraft, and craters to the teens for minimums, but a 3.5 GHz dual core/four thread Kaby Lake CPU has no problem passing 40 FPS? I'm skeptical, to say the least. And if memory bandwidth was the issue, then I'd expect the UHD 630 to tank even worse, since it's running with slower memory than they tested the 2400G and 2200G with.

Ryzen is ~ Sandy Bridge in most games, some slightly higher, some much lower.

2200/2400G are simply worse Ryzens in multiple ways.

This really shouldn't need to be rehashed every time the obvious happens again.




And don't compare Skylake IMC/Cache architecture with Ryzen.

AMD is simply licensing the IMC they are using in Ryzen.

Skylake IMC/Cache architecture is the best in the world.

To be clear, I'm saying that G4560 would beat Any Ryzen variant clocked at 4.2 Ghz in that particular test in CPU limited areas.

The fact that this particular Ryzen also has a parasitic client to the only memory interface it has means it's obviously going to lower that ideal situation (of having no memory controller load from a parasitic second source) even further.

 

[Red Hawk]

Ryzen is ~ Sandy Bridge in most games, some slightly higher, some much lower.

2200/2400G are simply worse Ryzens in multiple ways.

This really shouldn't need to be rehashed every time the obvious happens again.

And don't compare Skylake IMC/Cache architecture with Ryzen.

AMD is simply licensing the IMC they are using in Ryzen.

Skylake IMC/Cache architecture is the best in the world.

Ryzen is just so bad that it can't do more than 30 FPS in Starcraft 2? That's really the what you're going with as "the obvious happening"? With that, it's pretty clear you're not interested in having an honest discussion, because come the f*dge on.

https://www.techporn.ph/amd-ryzen-7-1800x-am4-cpu-review/

Even granting that Ryzen has significantly weaker single threaded performance than Skylake -- it should be able to do better than thirty-flipping-frames per second.

 

[24601]

Ryzen is just so bad that it can't do more than 30 FPS in Starcraft 2? That's really the what you're going with as "the obvious happening"? With that, it's pretty clear you're not interested in having an honest discussion, because come the f*dge on.

https://www.techporn.ph/amd-ryzen-7-1800x-am4-cpu-review/

Even granting that Ryzen has significantly weaker single threaded performance than Skylake -- it should be able to do better than thirty-flipping-frames per second.

You can make a test do anything you want in Starcraft 2.

If you just idle on the screen with zero CPU action like that test you are linking obviously did then you will end up with a purely draw call limited "benchmark" instead.

The benchmark that the other site is using is during actual gameplay, and the fact is obvious when you see that the AMD and Nvidia GPU are CPU bottlenecked at the same FPS, which means that that particular benchmark is testing a non-draw call limited workload, but instead a different CPU limited workload (which there are plenty to choose from in a real RTS).

 

[24601]

It's extremely obvious to me what a particular Starcraft 2 benchmark is doing for me, since I played the game a very long time and know exactly what the site you are linking to is doing to "juice" Ryzen's appearance in their "benchmark"

If you are getting over 100 fps at max settings then you are doing basically nothing other than staring at a blank screen.

 

[Red Hawk]

You can make a test do anything you want in Starcraft 2.

If you just idle on the screen with zero CPU action like that test you are linking obviously did then you will end up with a purely draw call limited "benchmark" instead.

The benchmark that the other site is using is during actual gameplay, and the fact is obvious when you see that the AMD and Nvidia GPU are CPU bottlenecked at the same FPS, which means that that particular benchmark is testing a non-draw call limited workload, but instead a different CPU limited workload (which there are plenty to choose from in a real RTS).

If what you're saying is true, benchmarks of Starcraft 2 should exist for Ryzen processors that exhibit the same behavior.

Can you come up with such benchmarks?

 

[24601]

If what you're saying is true, benchmarks of Starcraft 2 should exist for Ryzen processors that exhibit the same behavior.

Can you come up with such benchmarks?

When did you stop beating your wife?

I gifted my time to spoon-feed you in every single one of your inane questions and the only thing you know how to do is complain that you don't like the answers to the begging that you keep doing.

There are no questions in a "discussion" or "debate".




Insulting and berating other members is simply not allowed.


esquared
Anandtech Forum Director

 

[Red Hawk]

When did you stop beating your wife?

Yeah, uh huh, if you think it's a loaded question go ahead and disprove my assertion.

 

[alexruiz]

The 1030 and Vega 11 maybe close to each other in several games, but the 1030 absolutely trounces the Vega 11 in several other titles!

https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,11
https://www.purepc.pl/procesory/tes...ven_ridge_zen_i_vega_w_jednym_ciele?page=0,15

You forgot this one:

I am honestly amazed at how dense, biased or both some of you can be.
Reviews across the web show the 2400g on par with the RX 550 / GT 1030 in graphics performance. The consensus by most was in fact that the 2400g was slightly faster (techpowerup, pcworld, extremetech, guru3d)
In CPU performance, it was faster across the board than the Ryzen 5 1400 as a similar price point.

Yes, there are the outlier results, and for each one that goes one way, we can find another one going the other way. Hey, the A12-9800 is faster than a GT 1030, we have graphs to prove it! (sarcasm)
Results have to be read, understood and analyzed.
BTW, PCLab are also really dense, really biased or both. Their results on AMD hardware are always consistently lower than the rest, so these guys have no clue on how to configure AMD systems.
Maybe they just deploy from an intel image with a meltdown patch to amd hardware...

 

[24601]

Yeah, uh huh, if you think it's a loaded question go ahead and disprove my assertion.

I have spoonfed you 11 times in a row now.

Do you even know what a "discussion" or "debate" is?

Why do you keep stating rhetorical statements disguised as questions?

To what degree do you lack the ability to present an actual argument instead of begging people to spoonfeed you?

Can't answer?

Guess you are simply wasting my time then?

Why do you keep getting offended when I spoonfeed you the things you clearly do not know anything about?





Insulting and berating other members is simply not allowed.


esquared
Anandtech Forum Director