AMD Ryzen 7 1800X 3.6 GHz

[qu4k3r]

Good job!

I'd like to see 1700 review paired with B350 motherboard.

 

[SKD007]

Review would have been more helpful if it was 1440p and 4k instead of 1080p..
any who pays $499 for a processor will have a decent graphics card that can run 4k or 1440..

even if they don't have 4k monitor. They can still run dynamic resolution.

I run games at 1800p with 280x tweaked and now waiting for mitx x300 or x370...

the wait is killing me

 

[G33k2Fr34k]

Looks like the 1800x is throttling in games:

 

[Kanan]

Looks like the 1800x is throttling in games:

No it's simply not using all 8 cores in gaming.

Review is generally okay, but has some mistakes here and there that add up to give me a mixed feeling about the review (not the CPU, I've read reviews about it before anyway). They should've simply shipped him a Asus board and he'd probably given a higher score, because those boards simply work better and also support the higher ram frequencies he wanted. Simply put, he basically got the worst "highend" board there is atm with Asrock and especially Asus both better.

Why I have mixed feelings about the review:

On one hand you applaud a total energy efficiency win for AMD, and then you talk about it in the conclusion like it was a "draw" or "near-draw" vs Intel.

On one hand you say the Ram isn't overclockable to 2933 MHz, and then you say you bought a new Ram and it worked, just to say again that it didn't make it to 2933. Strange.

That games only support up to 4 cores is only partially true atm (many support more cores) and will change in the near future, but you denied that by stating it will not. That's just a opinion on your side and I'm pretty sure a wrong one, but you're stating it like it's a fact, but it's not.

CCX problem's, latency stuff etc. not part of this review.

etc. as I said the review is "okay", but not a inch more than that.

 

[G33k2Fr34k]

No it's simply not using all 8 cores in gaming.

Review is generally okay, but has some mistakes here and there that add up to give me a mixed feeling about the review (not the CPU, I've read reviews about it before anyway). They should've simply shipped him a Asus board and he'd probably given a higher score, because those boards simply work better and also support the higher ram frequencies he wanted. Simply put, he basically got the worst "highend" board there is atm with Asrock and especially Asus both better.

I see. I guess you're right. It looks like most games don't utilize 8 cores properly.

 

[Rahmat Sofyan]

Hmmm... 8.6 overall

 

[Patriot]

I see. I guess you're right. It looks like most games don't utilize 8 cores properly.

best part of the 1700... get an asus board, push a button in software and 3.8ghz stock volts and cooler for $330... oh and... mine likes 3200 ram just fine.... the difference between 2133 and 2666 is the same as 2666 to 3200. another 5fps give or take.

 

[FordGT90Concept]

I see. I guess you're right. It looks like most games don't utilize 8 cores properly.

Most games target quad cores without SMT. Ryzen has 8 cores with SMT so as that video shows, the 1700 is chillaxing while the 7600K is straining. Sure 1700 has a little lower framerate but you're talking 87 FPS instead of 90. Who cares? Start something like Handbrake that can use 16 threads and suddenly the 7600K looks pathetic.

I think the score of 8.6 is fair. Ryzen has compatibility problems right now which deserves a point docked. The remaining 0.4 can easily be attributed to the asterisks describing how some games take a significant performance hit on it and the relatively small L3 cache. Overall though, it's the best bang for the buck on the market once you get it running.

 

[ISI300]

Platform issues aside, seems like an amazing CPU and a great architecture to use as baseline for Zen+.
1700X is the one I'd go with if I eventually decide to ditch my 4670K.
Good job AMD, not very often we get to say that.

 

[notb]

Sure 1700 has a little lower framerate but you're talking 87 FPS instead of 90. Who cares? Start something like Handbrake that can use 16 threads and suddenly the 7600K looks pathetic.

This whole Ryzen situation - especially the comments under reviews/leaks - really made me wonder how people are using their PCs.
Seriously, how often do you encode movies?
Looking at the discussions, one could easily get an impression this is a rendering/encoding/physical simulation forum. Suddenly everyone praises the performance in tasks are - in general - performed by maybe 1% of PC users.

Also think about the qualitative nature of the benchmarks.
Obviously more cores give an advantage in multi-threaded tasks, but that's just in improved time. You can render/encode/simulate on pretty much any machine. Encoding a movie on Ryzen 1700 will take maybe 30% less time than on a similarly priced i7. Is this a big deal for a home user, who does that occasionally?
Gaming is so much different, as it always has a defined "usability" threshold for hardware. Difference of few fps in the >80 range is totally unimportant, but at ~20 it implies whether you can play the game or not.

For example: I'm not really into gaming lately - the list of titles I'd like to play is building up. For that reason I'm still using a very old desktop - with an almost ancient E5400. However, I'm doing a lot of simulations and of course I get the same results as on any other CPU - it just takes a few more hours.

It's very different with productivity tasks that you do "live" - like working with large Excel files. Here a poor CPU can be really frustrating. And yet, I find most reviewers have totally ignored this kind of benchmarks. For example the TPU review shows a result of an unknown Excel test that takes under a second. Maybe opening a file? This is not how professionals use Excel...

 

[RejZoR]

I'm gonna ask the obvious question, why is 6900K, a direct competitor to R7 1800X not listed in the charts? Comparing enthusiast level cruncher like 1800X to several mainstream Core i7's is a bit silly. Almost as if reviewer assumed only gamers would buy Ryzen... What about people who require crunching horsepower, but are not willing to pay 1100€ for 6900k ?

 

[notb]

I wanted to inform you personally that Ryzen isn't for you, due to lack of IGP. No man, you can't plug anything into the monitor outputs on the Ryzen motherboards! Ryzen APUs are coming out in H2. For the majority of people this is a non-issue, that's why it wasn't mentioned in the conclusion text.

I think it might not be an issue for typical TPU forum user, but as for majority of people it's actually pretty huge.
The reviews on TPU aren't read just by enthusiasts - the site is hugely popular and well positioned in search engines. Many people will just google for "Ryzen review" and find your text in top 3 results. They might not know much about CPUs and - after years of using Intel - be unaware that not every working computer offers a video signal.
As such, I would put the lack of IGP as the first con... probably in red bold font.

Also, could you share the specification for productivity tests? I'm mostly interested in the MS Office stuff, because the times shown suggest some very basic tasks. Weren't you tempted to use something more serious (some sites use a Monte Carlo benchmark for Excel).

 

[W1zzard]

I'm dumbfounded by the poor overclock performance. And the fact the chip actually looses performance in some tests.

I mentioned it somewhere. Due to XFR boosting single-thread workloads to 4.1, which is higher than 4.0 manual OC which is all cores though. So single thread apps will be worse with 4.0 manual OC, while multi-threaded ones will be better. The big difference is that XFR works out of the box, without manual tuning, at stock voltage.

I don't agree with you regarding integrated graphics being a con. A con because Intel offers it for nothing more than marketing? Anyone buying a 7700k to use it with the integrated graphics is completely throwing their money in the wrong direction.

It is a small con but worth mentioning in my opinion, if only to educate less experienced users, who might look at motherboards, their outputs and think "oh good it has integrated". For many office, workstation, render farm, servers, scientific, work tasks igp is good enough, and it comes free and integrated on non-HEDT Intel CPUs. Which also helps with formfactor. I know most of you will use a dedicated graphics card. Wouldn't be the first time a graphics card fails and IGP can be used while waiting for the RMA to complete, to at least be able to browse the web, do email and watch YouTube.

So nobody saw the 7770K typo yet...besides me?

Fixed. These charts are generated from a single entry in the database where I had a typo. Amazing indeed that nobody else noticed it (including our staff and proofreader)

does this benchmark reflect High Performance profile enabled?

Yes, enabled for all Ryzen testing

Also, could you share the specification for productivity tests? I'm mostly interested in the MS Office stuff, because the times shown suggest some very basic tasks. Weren't you tempted to use something more serious (some sites use a Monte Carlo benchmark for Excel).

I don't think anyone serious in the real world does Monte Carlo in Excel. You sound like you work in a workstation/science environment, tell me about your tasks (pm is fine ofc). I'd be happy to design and add more benchmarks

 

[Vlada011]

AMD Turbo work same as Intel Turbo?
I mean all processors are 100% stable on 4.0GHz clock but they could be overclock only 200-300MHz more, correct?

 

[FordGT90Concept]

Seriously, how often do you encode movies?

My server? Whenever necessary.

Encoding a movie on Ryzen 1700 will take maybe 30% less time than on a similarly priced i7.

If that 30% is the difference between realtime and buffering, it is worth it.

Difference of few fps in the >80 range is totally unimportant, but at ~20 it implies whether you can play the game or not.

If you're only getting 20 FPS and your computer isn't a dinosaur, consider upgrading the graphics card over the, well, everything. Witcher 3 was only using ~33% of the 1700 which translates to 5-6 of the 1700's logical cores. The bottleneck wasn't the CPU, it was the GPU or the applications inability to use more cores. We already have a two pronged solution to that problem: 1) Ryzen itself which makes 8-core processors affordable to consumers and 2) next gen APIs like D3D12 and Vulkan which gain frames largely because they divert as much work away from the CPU as possible in addition to multithreading it.

TL;DR: no gamer should not consider a Ryzen 8-core simply because Intel's quad-core is a little bit faster. Developers have been coding for quad-cores for half a decade now. They're just as ready as I am to move to reasonable six and eight core machines.

It's very different with productivity tasks that you do "live" - like working with large Excel files. Here a poor CPU can be really frustrating. And yet, I find most reviewers have totally ignored this kind of benchmarks. For example the TPU review shows a result of an unknown Excel test that takes under a second. Maybe opening a file? This is not how professionals use Excel...

If you're seeing poor performance using Excel, the problem is that you're using Excel. Spreadsheets < databases.

 

[Tsukiyomi91]

Finally a good review. AMD did a good job at the power consumption side, which is good. Kinda sad to see it can't push beyond 4.3GHz...

 

[W1zzard]

It's very different with productivity tasks that you do "live" - like working with large Excel files. Here a poor CPU can be really frustrating. And yet, I find most reviewers have totally ignored this kind of benchmarks. For example the TPU review shows a result of an unknown Excel test that takes under a second. Maybe opening a file? This is not how professionals use Excel...

The value is the average time for various tasks. In a workbook with 240k cells, also copying to another one with 10k cells. What do you do in Excel? Send me an example, w1zzard@techpowerup.com

 

[geon2k2]

The value is the average time for various tasks. In a workbook with 240k cells, also copying to another one with 10k cells. What do you do in Excel? Send me an example, w1zzard@techpowerup.com

Office stuff hasn't been an issue for 10 years by now, for productivity the best breakthrough was the SSD, even very old CPU's can handle excel just fine for most of the people. If you speak about databases and access ... then that is a different story, but even then I'd look at a faster SSD first

What I really want to see is 1CCX, 4 core, 8 threads gaming benchmark pretty please
I'm sure gaming performance will be better if all the cores used will be within the same CCX.

 

[efikkan]

How can a more expensive processor + motherboard offer more value?

I think you should check closer, you might be mixing it up:
Ryzen 1800X: $499
i7-6800K: $434

Looks like the 1800x is throttling in games:

It's called stalling, not trottling. Trottling would be the clock dropping.

 

[GreiverBlade]

well that went pretty much as expected
so we need to buy a 120 dollar kit of ram to compete with a cpu thats 160 dollars cheaper
deal of the century

once again amd hedged there bets on having more cores and not a single developer cares

just in case ... you know the counterpart of the 7700K is the R7 1700, which cost the same price and not 1800X ... and all gaming video review, at 1080p, put the 3.7-3.8ghz 1700 on par, +/-10-15fps to a 5.0ghz 7700K? Yes i know ... it's still 8C/16T versus 4C/8T ... but hey ... it's Intel's fault if devs don't give a damn about "above 4 core" (worth nothing but frequencies wise ... 3.7ghz vs 5.0ghz make it 1.3ghz slower but still keep up and even in "4thread only" games )

in my case if i sell my 6600K my Z170X Gaming 7 and my 4x4 2800 C14, i bet i would have enough to get a full R7 1700 system, my initial budget was for a 7700K alone to replace my 6600K

as i not only do gaming but sometime encoding and other task where Ryzen can shine ... there is no better option for the price asked.

well since i already have the fund for the CPU i just need fund for RAM and Mobo ... and i have a CPU Mobo and RAM to sell to fund it, now ... since Ryzen Mobo aren't somewhat more expensive than the Intel one ... and still offer the same advantages
i could even get some money back if i wait for the R5 1600 ( awww shoot ... and getting only 6C/12T instead of 4C/4T at the same price)

also on core count ... actually now it's 8 real core + SMT ... and not 4x2 CMT, single thread performance is not as high as it could have .... but still enough.

as the review say : "AMD is competitive again" ... and damn yes it is.

TL;DR: no gamer should not consider a Ryzen 8-core simply because Intel's quad-core is a little bit faster. Developers have been coding for quad-cores for half a decade now. They're just as ready as I am to move to reasonable six and eight core machines.

actually ... nope ... if a 7700K is a good idea then a R7 1700 is also a good idea (tho sometime you need to put HT off on a 7700K ... ) and yep a Intel is a "little" bit faster from what i saw ... nothing to fret over as long as playing in 1440p and up (or 1080p ... +/-10-15fps is not that high ... ok 20 ... on a select set of game that i never had/liked and would never touch )

all in all gamer should consider Ryzen's offer around the R5 1600/R7 1700 if they have a budget for a 6600/7600/6700/7700K ... even if devs coded games for 4 core for over a decade (not all ... mmo and RTS ... would probably benefit from Ryzen) but if they move on (thanks to AMD probably) well ... they can be ready.

my end word would be: waiting on the issues to get sorted, mobo and BIOS to get better and holding until the R5 line get out and get benched is not a bad idea, but for now

edit... i noticed that the R7 1800X consume less than a 7700K only now ... yep ... good job AMD, now i wonder about the 1700 power consumption

You forgot to mention if you're testing these games in DX11 or DX12 (at least the ones that support it).
It's specially the DX12 mode of games that are not running too well on Ryzen. Being a low level API, these games will need optimization to run well on Ryzen.
I already posed about this on TR forum. ComputerBase tested all games in DX11 and DX12. The difference is noticeable.

https://techreport.com/forums/viewtopic.php?f=2&t=119280

Ryzen easily looses 20-30% of performance in DX12 in many games.

7700K vs 1800X:

• 
  Battlefield 1 DX11: 116 vs 122
  Battlefield 1 DX12: 127 vs 90
  Deus Ex: MD DX11: 87 vs 80
  Deus Ex: MD DX12: 83 vs 63
  RotTR DX11: 152 vs 135
  RotTR DX12: 168 vs 117
  Total War: W DX11: 43 vs 40
  Total War: W DX12: 42 vs 30

actually DX12 games are not "ready now" they loose FPS even on Intel System ... and not many games are DX12 as for ROTTR ... 152fps in DX11 and 168fps in DX12 ... my 6600K with a 1070 does the opposite ... (aka: it loose FPS and not gain 16fps ) at almost the same ratio of the 1800X

DX12 need to be optimized before being a thing ... actually DX12 feel like a beta to me ... (Vulkan on the other hand ... )

 

[efikkan]

TL;DR: no gamer should not consider a Ryzen 8-core simply because Intel's quad-core is a little bit faster. Developers have been coding for quad-cores for half a decade now. They're just as ready as I am to move to reasonable six and eight core machines.

You know better than this.

The problem with Ryzen is that 1800X only achieves great gains in certain more unusual workloads, while the cheaper i7-6800K would be better at gaming, Photoshop, web browsing, among other things. Buying a product that's super good at 7 applications you never/rarely use is pointless, real world performance is the only thing that matters. We'll have to see if Zen+ contains a better prefetcher to feed it's huge computational throughput.

The "lack of" multithreading in games is not due to quad cores being mainstream for so long, but rather the inability for rendering to scale over multiple cores. Rendering is done in a pipeline, and the CPU is really just building a queue for the driver to feed the GPU. Even though the GPU workload is massively parallel, the creation of the queue is not. Splitting up the queue would be limited to separate rendering stages, like physics/particle simulation, etc. So there would be limited gains here. Granted, Direct3D 12 allows you to have multiple threads working on a queue, but then you'll need to synchronize them which will create more latency than gains, so it's simply pointless. But the creation of the queue is really not the problem here, it's all the rendering thread does besides creating the queue. Typically a game engine iterates a list of objects, and then invokes a render() on each one, which in turn invokes several functions to render each object separately. In these cases each object would result in 2 or more cache misses (each resulting in ~250 clocks wasted), which will result in >95% of CPU cycles stalling while waiting for data. Intel has a better prefetcher so it's able to mitigate these problems to some extent, but the only way to make the game scalable in terms of CPU performance is to make the code cache/branch optimized, which requires a complete redesign of an engine.

 

[medi01]

eople saying W1zzard is Bias need to review their history.

Well, this has been called out on reddit:

note that AMD has stated that mem speed defines "infinity fabric" (CCX to CCX connection) speed.

 

[notb]

My server? Whenever necessary.

This is not an answer to my question. How often, honestly? How many hours of material a month? How much does it take? Can you estimate it?

If that 30% is the difference between realtime and buffering, it is worth it.

Nope. Real life shows that, for example, if you always run some tasks during sleep or while you're at work, 4 hours is just as good as 6.

If you're only getting 20 FPS and your computer isn't a dinosaur, consider upgrading the graphics card over the, well, everything.

Hmmm... At the moment I don't have a dGPU (occasional gaming on IGP), so replacing the card would be fairly difficult.
So as you can see for me Ryzen is really bad as a gaming CPU. I would get 0 fps in games.

But jokes aside: the argument remains valid in my opinion: IMO people are too concerned about benchmarks covering rare tasks.
I don't think a typical user is so affected by having to spend extra 30% time encoding movies or zipping files. However, we're all very much relying on web performance and this kind of tests are, IMO, heavily underestimated. I've seen reviews without them. This will be very important for lower Ryzen chips - especially APU.
Good multi-thread optimization could make them fine for gaming (for a long time), but potentially poor performance in single-thread tasks will surface at some point in the future (and some people don't replace their CPUs for half a decade or more).

Developers have been coding for quad-cores for half a decade now. They're just as ready as I am to move to reasonable six and eight core machines.

I'm seeing this theory all the time. Why do you assume game designers will optimize for 6-8 cores now? What new does Ryzen bring us when AMD has been selling 8 core consumer chips (actually cheaper than Ryzen 7) for years?
It's even worse today, because so many people moved to gaming laptops. Will AMD give us an 8c mobile Ryzen or just an 4c APU?
And if it an 8c Ryzen, what will be the battery life (as it will have to rely on dGPU for everything)?
Laptops with Intel CPUs can switch between IGP and dGPU, so they can be used as everyday notebooks or workstations (if you can live with the design/build quality).


If you're seeing poor performance using Excel, the problem is that you're using Excel. Spreadsheets < databases.[/QUOTE]

 

[rtwjunkie]

Z

Looks like the 1800x is throttling in games:

It looks to me like gaming is not working it hard enough. It needs coding for moar cores by developers.

 

[MrMilli]

actually DX12 games are not "ready now" they loose FPS even on Intel System ... and not many games are DX12 as for ROTTR ... 152fps in DX11 and 168fps in DX12 ... my 6600K with a 1070 does the opposite ... (aka: it loose FPS and not gain 16fps ) at almost the same ration of the 1800X

DX12 need to be optimized before being a thing ... actually DX12 feel like a beta to me ... (Vulkan on the other hand ... )

Yes, but Ryzen loses much more performance compared to Intel in DX12. The problem is that most reviewers solely use DX12 for testing, showing Ryzen in a worse light than needed.

Look at the FPS difference again: https://techreport.com/forums/viewtopic.php?f=2&t=119280