Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[LMF5000]

Those shenanigans tend to be an Intel only endeavour.

AMD doesn't seem to cut down the feature-set when binning or offering fused parts.

[I assume that is what you mean? As in, will they all support AMD-V, ECC memory etc]

Thanks all for the replies. My question was in terms of transistors on the silicon - do the 1700X amd 1800X come from the same wafer or are they dedicated, different designs? Or perhaps the lower spec models have damaged parts of the die disabled to improve yield, same as nvidia does with some lower-binned parts that actually come from the same wafer as the more expensive products.

I plan to build one for a work PC. The budget is quite flexible, but the 1800X is €150 more expensive than the 1700X and only 5-8% better in compute benchmarks. If the difference is just clock speed I could try and imitate a 1800X by overclocking. For motherboard I was considering the cheapest Asus AM4 model (the Prime B350M-A) with 4 sticks of 4GB 3200MHz DDR4 (whichever brand is in stock at the time).

 

[IEC]

All the Ryzen 7 chips contain exactly the same # of cores, CCX configuration, and L1/L2/L3 caches.

Ryzen 5 chips will be the lower binned 6 and 4 core harvested parts (but again, physically the same prior to harvesting)

 

[Mopetar]

Can someone explain to me why Ryzen performs better at 1440p versus 1080p in gaming? I have been seeing this in a lot of reviews lately and it seems that Ryzen performs better when there's less stress on the CPU.

http://www.relaxedtech.com/reviews/amd/ryzen-7-1700/7

For some games, performance doesn't improve with more cores and Ryzen can't reach nearly the same clock speed as an i7 4C/8T chip so even though the IPC is close it doesn't matter.

As resolution increases, more stress is put on the GPU and eventually it becomes the limiting factor. It's like having a really fast car but always being stuck in traffic for the CPU so it doesn't matter how good/bad the CPU is at that point. An old Geo Metro and a new Porsche both got the same speed in gridlock traffic.

That's why at 4K resolution even a cheap Pentium is as good as a $1K HEDT i7 because both can run faster than the GPU requires.

 

[CatMerc]

So, what Dx11 has to do with bad geometry throughput and general front-end performance, that as such limits access to vast overkill of ALU array GCN cards boast? I guess that's the time the old saying about bad dancer and his balls applies.


https://www.computerbase.de/2017-03.../4/#diagramm-rise-of-the-tomb-raider-dx12-fps
https://www.computerbase.de/2017-03...4/#diagramm-ashes-of-the-singularity-dx12-fps
http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,16.html
http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,17.html
http://www.guru3d.com/articles_pages/amd_ryzen_7_1800x_processor_review,19.html

All right, so here's the deal. We couldn't find the testing notes, only results, so we're redoing testing. As much as I trust my memory, it's a lot better to have actual (or digital) paper tell you exactly what you did and your results. This time to ensure no GPU bottlenecks, we're dropping down even further to 480p.

Thanks for pointing out the oddity!

 

[darkswordsman17]

PSA - Microcenter offering $100 off on Ryzen 1700 when bundled with AM4 motherboard. Only in store. And Yes, you can thank me

http://www.microcenter.com/product/..._AM4_Boxed_Processor_with_Wraith_Spire_Cooler

Regards,
Vishnu

It's saying only $50 off?

Actually in just the past few minutes the R 1700 dropped to just $30 off, but the 1700X/1800X are $50. Still shows $50 on the page for the 1700, but this page shows $30 (was showing $50 just a few minutes ago).

http://www.microcenter.com/site/products/amd_bundles.aspx

D'oh, in store (assumed their website would list it just say you have to buy in store but I assume that only shows in store?)

 

[Malogeek]

It's saying only $50 off?

Yeah. It was showing $100 off before though.

 

[IEC]

Yeah. It was showing $100 off before though.

Someone goofed. The discount was supposedly for the 1700X/1800X, but a few people got MC to honor the discount for the 1700. Seems like it's $50 (X models) / $30 (1700) now, however.

 

[AMDisTheBEST]

I am waiting to see how Zen+ next year performs. If AMD fixes the shortcoming in gaming while bumping up the clock speed, they won't even need to raise ipc further to compete.

 

[SunburstLP]

Can someone explain to me why Ryzen performs better at 1440p versus 1080p in gaming? I have been seeing this in a lot of reviews lately and it seems that Ryzen performs better when there's less stress on the CPU.

http://www.relaxedtech.com/reviews/amd/ryzen-7-1700/7

I've noticed the same, but I haven't drawn a conclusion yet. My gut says it's an oddity from NV drivers. I don't think they had access to samples to make any optimizations. It's certainly in their best interest to optimize. The better their cards look, the more they sell.

 

[Topweasel]

I am waiting to see how Zen+ next year performs. If AMD fixes the shortcoming in gaming while bumping up the clock speed, they won't even need to raise ipc further to compete.

Zen plus won't be needed. I have been digging into the benchmarks and reviewing all of the different tests. Vega will fix Ryzen. It will force Nvidia to fix the DX12 numbers. If you really analyze the numbers Nvidia drivers stop scaling on DX12 after 4c. Vega if even 90% as good in DX11 as the 1080, it will destroy the Titan in DX12. Since it it optimized for 7700 right now at the cost of the 6900, 6950 and R7, Nvidia will have to fix the driver in hopes to keep up (and probably advertising the 6900k and Titan X as the best combination ever). All of a sudden the benches that make Ryzen look waay off in the reviews will get reversed.

 

[sushukka]

For some games, performance doesn't improve with more cores and Ryzen can't reach nearly the same clock speed as an i7 4C/8T chip so even though the IPC is close it doesn't matter.

As resolution increases, more stress is put on the GPU and eventually it becomes the limiting factor. It's like having a really fast car but always being stuck in traffic for the CPU so it doesn't matter how good/bad the CPU is at that point. An old Geo Metro and a new Porsche both got the same speed in gridlock traffic.

That's why at 4K resolution even a cheap Pentium is as good as a $1K HEDT i7 because both can run faster than the GPU requires.

Car analogies ftw. And when you compare Intel 4c desktop processors with something else than gaming load, Ryzen is like a truck compared to a car. Maybe the parcel don't move that fast but the amount of goods moved at the same time matters.

 

[imported_jjj]

Apparently Total War Warhammer got an update last week that helps a bit with Ryzen http://www.hardware.fr/news/15076/maj-total-war-warhammer-ryzen.html

 

[imported_jjj]

Ryzen 5 1600 review (in Spanish) https://elchapuzasinformatico.com/2017/04/amd-ryzen-5-1600-review/

 

[guachi]

Thanks for the link. The Google auto translate works quite well with Spanish, aside from the silliness of calling it a 6 core/12 threads processor.

EDIT: According to those benchmarks, the 1700X is 22.7% to 45.6% faster in multi-threaded benchmarks compared to the equivalent frequency 1600.

The 1600 at $220 might be the sweet spot for people who might use many cores occasionally (i.e., more gaming than encoding, for example). The 1700 is 50% more expensive (and can probably be clocked equal to the 1600, I'd think) for 50% more cores/threads but is just shy of 50% faster in multi-threaded applications (and can be super-efficient at stock) but isn't really much faster in games. The only game I'd think the 8-core CPU would be really good in is BF1 multi-player from the benchmarks and first-hand play accounts I've read.

 

[thepaleobiker]

Ryzen 5 1600 review (in Spanish) https://elchapuzasinformatico.com/2017/04/amd-ryzen-5-1600-review/

Interesting to note how the R5 beats the R7 1700 in Doom OpenGL as well as Tomb Raider Dx11 benches.

 

[imported_jjj]

Thanks for the link. The Google auto translate works quite well with Spanish, aside from the silliness of calling it a 6 core/12 threads processor.

EDIT: According to those benchmarks, the 1700X is 22.7% to 45.6% faster in multi-threaded benchmarks compared to the equivalent frequency 1600.

The 1600 at $220 might be the sweet spot for people who might use many cores occasionally (i.e., more gaming than encoding, for example). The 1700 is 50% more expensive (and can probably be clocked equal to the 1600, I'd think) for 50% more cores/threads but is just shy of 50% faster in multi-threaded applications (and can be super-efficient at stock) but isn't really much faster in games. The only game I'd think the 8-core CPU would be really good in is BF1 multi-player from the benchmarks and first-hand play accounts I've read.

33% more cores and for valuation, depends how you look at it
If you look at CPU vs CPU prices, it is 50% more but if you look at all parts purchased, it's at least CPU+mobo but can the entire system. So if you include a 120$ mobo and up, you get bellow a 33% premium.

Interesting to note how the R5 beats the R7 1700 in Doom OpenGL as well as Tomb Raider Dx11 benches.

They get 5.7ns lower memory latency with the 1600 for some reason so maybe the results for the 1700 aren't done with the current BIOS.

 

[Puffnstuff]

Probably memory latency rather than bandwidth is the reason.

There is a direct relationship between the DDR speeds and the CCX as the cache bandwidth has been increased using the higher speed memory. It would be foolish to build a Ryzen system and skimp using cheap low speed memory.

 

[thepaleobiker]

There is a direct relationship between the DDR speeds and the CCX as the cache bandwidth has been increased using the higher speed memory. It would be foolish to build a Ryzen system and skimp using cheap low speed memory.

True. All the reviews and AT forum reports point to Ryzen & its CCX loving fast ram speeds, with little penalty using lax timings (3000 DDR4 CL16 is > 2666 DDR4 CL14 : not sure if these timings exist but you get the idea)

 

[unseenmorbidity]

Apparently Total War Warhammer got an update last week that helps a bit with Ryzen http://www.hardware.fr/news/15076/maj-total-war-warhammer-ryzen.html

Interesting. It's not faster than the 6900k in that game.

 

[Doom2pro]

Per HwInfo latest release notes, it looks like AMD is using the Thread Ripper name afterall...

"Improved support of AMD ThreadRipper"

 

[CatMerc]

Per HwInfo latest release notes, it looks like AMD is using the Thread Ripper name afterall...

"Improved support of AMD ThreadRipper"

PREPARE FOR AMD'S THRRRRRRRREAAAAAAAAAAAADDDDDRRRRRRRRRRIPPPPPPPPPPPPPPPPEEEERRRRRRRRRRRRRR TM 32 CORES OF PURE THREAD DECIMATING GLORY! Don't buy one for your girlfriend, or she'll end up pregnant! AMD's THREADRIPPER TM is a monster truck of computing! Watch it run over UP TO 64 THREADS while you buy a $9 beer! THREADRIPPER.
*AMD ThreadRipper may only be used by end-users wearing Affliction or Tap-out T-shirts

 

[guachi]

33% more cores and for valuation, depends how you look at it

Well, duh. I can do math! Yes. It's 33% more cores for 50% more dollars (compared to the 1700)

 

[unseenmorbidity]

Have you guys seen this strange result?

 

[PPB]

I wouldnt trust elchapuzas and i0 for that matter. Shoddy testing methodology.

Sent from my XT1040 using Tapatalk

 

[coercitiv]

Have you guys seen this strange result?

As long as we don't have confirmation both R5 and R7 were using same board & same BIOS, it's not strange at all.

In fact, judging by their previous 1700X review (image on the right) the results in their current review are identical to those collected 1 month ago.