AMD Ryzen 1700X CPU Review @ [H]

Investors seem to shy away from AMD stock after a lot of the reviewers reported lower gaming results. So at least the idea of gaming capability matters to some, and possibly why Lisa Su addressed it.

Shares of AMD fall after gaming performance of new processors disappoints.

"Advanced Micro Devices shares dipped on Thursday after the launch of their new Ryzen line of desktop CPUs disappointed, with some technology reviewers unhappy about the poor gaming performance."

http://www.cnbc.com/2017/03/02/shar...pu-disappoints-with-gaming-performance-.html?

AMD always had that coming. They expect shining brilliance out of the gate from the weakest competitor.
 
That problem with that is its only happening in games, why? This is not API related, cause if it was it would show up in anything that uses DX or Vulkan etc. So now OS related, probably not, cause AMD has had quite a bit of time to tell MS what needs to be done, its not done in a closed box, nor will AMD not send MS samples to ensure things are running well. App specific related, well its only happening in games, I would think it would happen else were too if it was programming side.

Nope its something to do with Ryzen and its architecture.
I think it's too early to tell. If another SKU gets released and magically the problem is fixed then you'll have your answer.
 
Investors seem to shy away from AMD stock after a lot of the reviewers reported lower gaming results. So at least the idea of gaming capability matters to some, and possibly why Lisa Su addressed it.

Shares of AMD fall after gaming performance of new processors disappoints.

"Advanced Micro Devices shares dipped on Thursday after the launch of their new Ryzen line of desktop CPUs disappointed, with some technology reviewers unhappy about the poor gaming performance."

http://www.cnbc.com/2017/03/02/shar...pu-disappoints-with-gaming-performance-.html?


Bound to happen when AMD targets gaming enthusiasts and doesn't deliver. They sent the wrong message.
 
Yeah, it is clear that something is wrong.

In single threaded Cinebench Ryzen is pretty much matching a Skylake i5-6600K, so we know the thread performance is there. It has twice the cores and 4 times the threads of a 6600k though, and while more cores don't always add to the performance in games, they never - in my experience - hurt.

This is a brand new architecture, and this is something we haven't seen in some time. Intel's designs have used incremental improvements on previous designs for decades. It is only to be expected that there will be some teething problems for a brand new from scratch architecture.

It sounds to me like the Windows scheduler isn't fully optimized for handling Ryzen's SMT setup yet. We also know that there are some BIOS patches coming down the pike to deal with - among other things - RAM clock issues, which we are told currently only hit 2133 (accurate? I'd have to re-read the review) but should be able to hit 3000+.

So, over the next couple of months, I think we'll see OS patches, driver updates and BIOS updates that will improve the situation. The question will be, just HOW much of a difference will these result in.
Well for sure it will make motherboard reviews way more interesting and make Dan work much harder but probably way more fun for him as well. Will make for some interesting revelations and hopefully some great surprises as time goes on. I think through the reviews here we will see the progress or not. New game reviews may take a whole new meaning when you have an Intel platform and an AMD one and not just video card differences if possible.

Speaking of games, will we see a top three platform game review with the 1080Ti? x99, Z270 and X370? Seeing if max gaming IQ and resolution really makes hill of beans in the end which platform you game on. What Kyle did for VR was extraordinary with the number of titles reviewed? Must not have sleep in a week.
 
Last edited:
Investors seem to shy away from AMD stock after a lot of the reviewers reported lower gaming results. So at least the idea of gaming capability matters to some, and possibly why Lisa Su addressed it.

Shares of AMD fall after gaming performance of new processors disappoints.

"Advanced Micro Devices shares dipped on Thursday after the launch of their new Ryzen line of desktop CPUs disappointed, with some technology reviewers unhappy about the poor gaming performance."

http://www.cnbc.com/2017/03/02/shar...pu-disappoints-with-gaming-performance-.html?

Poor gaming performance may be an overstatement. I still have a feeling sales will be strong and AMD has nothing to worry about.
 
Nope its something to do with Ryzen and its architecture.

It could very well be Architecture related. In fact, I'd expect it to be. Ryzen is not an Intel architecture. Over the last 10 years game engines have gone after Intel optimizations and tweaking for them to get the most out of them. It shouldn't be a surprise that they have some significant optimization advantages. Some game engines may have optimized for the Bulldozer architecture as well, but guess what, Ryzen isn't the same architecture as Bulldozer either (thank god).

For existing titles, I'd expect this probably won't change. In some cases game devs go back and update their engine. TripWire did this for Red Orchestra 2 - as I recall - when they after launch upped the rev of the underlying Unreal engine, but I get the impression this is pretty rare. A game tends to stay on the rev engine it is launched.

Now, if Ryzen starts to get some notable market share, I think we will see engine makers and game devs pay it some more attention. What is interesting to me is that we have seen it scale very well in VR titles in Kyles testing. What's notable about VR titles is that they are all rather new, and built on top of rather new engines. This - to me - is a good sign of things to come.
 
Investors seem to shy away from AMD stock after a lot of the reviewers reported lower gaming results. So at least the idea of gaming capability matters to some, and possibly why Lisa Su addressed it.

Shares of AMD fall after gaming performance of new processors disappoints.

"Advanced Micro Devices shares dipped on Thursday after the launch of their new Ryzen line of desktop CPUs disappointed, with some technology reviewers unhappy about the poor gaming performance."

http://www.cnbc.com/2017/03/02/shar...pu-disappoints-with-gaming-performance-.html?

Exactly what I predicted would happen if these processors had luckluster gaming performance.
 
Any thoughts on a "frames per dollar" evaluation that might show the value of a CPU based on how many frames you get per dollar spent? Taking the [H] numbers and dividing by retail price (newegg/amazon) this is what i came up with:

CPU price lp uni bio metro aots
1700x 400 0.82 0.71 0.88 0.42 0.10
7700k 350 1.1 1.24 1.36 0.56 0.12
6900k 1050 0.46 0.34 0.44 0.22 0.05
2600k 280 1.078 1.13 1.27 0.57 0.11


This kind of shows that dollar-for-dollar the 7700k is probably the best performer (excluding the cost of the rest of the platform).
never understood that logic. for gaming and bdgeting. I get it for rendering and server stuff but not for FPS in games valuations. It either works or doesn;t work for gaming and for me minimum rate rate is most important which is why i run a fast 4 core
 
Why not wait to to see what the issue is? Other review sites clearly said it feels like a bug of some kind. Others got better performance if they disabled SMT. This was issue with hyper threading in early days so yea I have heard of this sort of stuff before.


Hyper threading issues was more then just games though, it was prevalent in many applications. Just can't draw parallelisms with that
 
Yeah wasn't that the one the crippled overclocking but improved stability? Or was that for sky lake this one is before. but still.

That had nothing to do with performance.

The point is microcode can alter performance. Think working around a prefetch bug, branch prediction error or preventing cache entries from expiring early. Unfortunately we don't know how many/what parameters can be twiddled with after the fact.
 
Did you not read the article? Overclocking puts the 1700 and 1800X around 4.0GHz to 4.1GHz max at around 1.4v.

I skimmed the article so that's my mistake. I thought those 4Ghz speeds were from boosting.
 
Hey, Did you forget us the better mankind on Linux ?........
I am deeply disappointed - Anyways, have you checked out phoronix for their results, no gaming there but that was quite good vs the 7700K in pretty much everything...



We may be missing a bit of performance but I assume we won't see ryzen surpassing intel in Gaming but closer to matching at best
for SMT - I'm still struggling with a Xeon 2680 V2 with SMT in some use case scenario and, Intel have had SMT for a gamer kids lifetime!
So it's not to be unexpected that SMT can be improved and improve results by upto 5-7% as some report.

I hope this is true for the future competition which I hope will go on.
The Ryzen CPU as is seems to be well beyond expectations really, who would have thought we'd see something from amd again that actually can compete and trade blows in.... anything :)

Agreed. I don't think we will see them surpassing Intel. We didn't expect that. I do suspect we will see them creeping closer over time as teething issues are worked out though.
 
That problem with that is its only happening in games, why? This is not API related, cause if it was it would show up in anything that uses DX or Vulkan etc. So now OS related, probably not, cause AMD has had quite a bit of time to tell MS what needs to be done, its not done in a closed box, nor will AMD not send MS samples to ensure things are running well. App specific related, well its only happening in games, I would think it would happen else were too if it was programming side.

Nope its something to do with Ryzen and its architecture.

That is a little dramatic, it is a new architecture that will take bleeding in, Bulldozer alone improved on initial benches once Windows updated and firmware was updated, in early tests there were issues of parking and I think there is some core parking going on. It is buggy but I don't think it is a "end of the world" for 5 years thing. 1440P is solid so problems work both ways, you should stop always taking the negative without even contemplating that problems are not always major. Synthetic wise the performance is better optimized.

I think people are going way overboard. I mean it beats a Sandybridge at 4.5Ghz, it is out of GPU bottleneck and can run high FPS to play over game line, it may not be the fastest but it is not like any sane person expected that. I made my position clear from the get go, it will perform like a stock clocks 5960X generally across the board and it is kind of at that level, very mainstream in gaming but enough if you can make use of the exceptional creation power of the CPU.
 
The point is microcode can alter performance. Think working around a prefetch bug, branch prediction error or preventing cache entries from expiring early. Unfortunately we don't know how many/what parameters can be twiddled with after the fact.

Oh I agree, but its not that much, and will only happen with small amounts of things. Its rare that we see it increase across the board.
 
Well for sure it will make motherboard reviews way more interesting and make Dan work much harder but probably way more fun for him as well. Will make for some interesting revelations and hopefully some great surprises as time goes on. I think through the reviews here we will see the progress or not. New game reviews may take a whole new meaning when you have an Intel platform and an AMD one and not just video card differences if possible.

Second that. If anything Dan's work over the next 6 months certainly got a lot more interesting.
 
Agreed. I don't think we will see them surpassing Intel. We didn't expect that. I do suspect we will see them creeping closer over time as teething issues are worked out though.

I would be surprised if a whole lot changed through updates besides memory clocks. BIOS updates don't normally lead to higher performance, OS updates could but I'm not convinced. Even if an update makes a difference I don't think it will close the gap completely, if much at all. Time will tell though.
 
Oh I agree, but its not that much, and will only happen with small amounts of things. Its rare that we see it increase across the board.

I won't deny being overly optimistic! ;) I've a creaky old 1090T on water that I have had to dial back to 3.8 from 4.1GHz and would really like to get more cores without having to spend Intel money. :nailbiting:
 
That is a little dramatic, it is a new architecture that will take bleeding in, Bulldozer alone improved on initial benches once Windows updated and firmware was updated, in early tests there were issues of parking and I think there is some core parking going on. It is buggy but I don't think it is a "end of the world" for 5 years thing. 1440P is solid so problems work both ways, you should stop always taking the negative without even contemplating that problems are not always major. Synthetic wise the performance is better optimized.

I think people are going way overboard. I mean it beats a Sandybridge at 4.5Ghz, it is out of GPU bottleneck and can run high FPS to play over game line, it may not be the fastest but it is not like any sane person expected that. I made my position clear from the get go, it will perform like a stock clocks 5960X generally across the board and it is kind of at that level, very mainstream in gaming but enough if you can make use of the exceptional creation power of the CPU.

Well that is the problem isn't it, that is not the way AMD marketed this product, they marketed as gaming enthusiast, overclock enthusiast products. Neither are true at this point. Its still at Sandy bridge level of gaming performance a 4 core part that is 6 generations old!
 
That problem with that is its only happening in games, why? This is not API related, cause if it was it would show up in anything that uses DX or Vulkan etc. So now OS related, probably not, cause AMD has had quite a bit of time to tell MS what needs to be done, its not done in a closed box, nor will AMD not send MS samples to ensure things are running well. App specific related, well its only happening in games, I would think it would happen else were too if it was programming side.

Nope its something to do with Ryzen and its architecture.
If that's the case what's going on in synthetics? You'd think they would suffer the same way? This is interesting and will certainly be a fun one to watch over the next several months to see exactly what's going on here, but something clearly isn't working as intended.
 
I'd be interested in seeing 4 and 6 core chips with higher clocks for gaming, might be more competitive with the 1150 i7s in gaming.

Otherwise pretty cool, a bit slower than i thought for gaming.

If AMD can come up with a competitive 4 core chip with higher clocks for gaming, they would have done so in their current line up. They know they can't match the IPC and the clocks, therefore they come up with this strategy of selling more cores to you at a cheaper price per core. Having said that, if AMD comes up with a 4 core chip with the same price per core, would people buy it for other purposes than gaming?
 
Last edited:
If AMD can come up with a 4 core chip with higher clocks for gaming, they would have done so in their current line up. They know they can't match the IPC and the clocks, therefore they come up with this strategy of selling more cores to you at a cheaper price per core.

Yep, its back to selling more slower cores. But that doesn't work out for gaming and regular tasks.
 
Poor gaming performance may be an overstatement. I still have a feeling sales will be strong and AMD has nothing to worry about.

AMD has plenty to worry about. Not that gaming would've made much of a difference to their bottom line, though. Plenty of sales to be made on the server market, that's for sure. When's their next financial call?
 
If that's the case what's going on in synthetics? You'd think they would suffer the same way? This is interesting and will certainly be a fun one to watch over the next several months to see exactly what's going on here, but something clearly isn't working as intended.

Didn't the first run of Phenoms have a TLB lookaside bug that was computationally expensive to workaround? Certain workloads would trigger it more reliably than others.
 
Well that is the problem isn't it, that is not the way AMD marketed this product, they marketed as gaming enthusiast, overclock enthusiast products. Neither are true at this point. Its still at Sandy bridge level of gaming performance a 4 core part that is 6 generations old!

Where are you getting this from??? They were touting RENDERING benchmarks and talk about going head to head with Broadwell-E....and they nailed that. I never got the impression this was going to go toe to toe with the fast quads. Where were you people?
 
Well, it's not quite a home run chip like we all hoped, but it's definitely a solid triple (too keep up the baseball analogy lol). For anyone doing things outside of thr gaming world, Ryzen looks like a pretty awesome value. Parity with Intel's $1k cpu in workstation type work for half the cost is no joke, and hopefully it will make Intel lower their prices on their HEDT lineup going forward.
 
Seems like there's a lot of kinks to work out for top notch performance from Ryzen.

Would love to see a "3 month" or "6 month" later review to see if much has changed as far as fixes and updates.


Given that this is brand new technology and the issue has already been identified, I would say this would be a good idea. Although it sounds like the gaming performance comes down to games being optimized for Intel's instruction set. So it may be worthwhile to do the recheck on games as those patches come available, if Dr. Su was correct.
 
Well for sure it will make motherboard reviews way more interesting and make Dan work much harder but probably way more fun for him as well. Will make for some interesting revelations and hopefully some great surprises as time goes on. I think through the reviews here we will see the progress or not. New game reviews may take a whole new meaning when you have an Intel platform and an AMD one and not just video card differences if possible.

Speaking of games, will we see a top three platform game review with the 1080Ti? x99, Z270 and X370? Seeing if max gaming IQ and resolution really makes hill of beans in the end which platform you game on. What Kyle did for VR was extraordinary with the number of titles reviewed? Must not have sleep in a week.

Second that. If anything Dan's work over the next 6 months certainly got a lot more interesting.

I definitely have a lot more work to look forward to. The variety and learning a new platform increases the difficulty but it definitely increases the amount of enjoyment I'll get from it. I'm very excited to get ahold of some AM4 motherboards. You can only look at so many X99 or Z270 motherboards before they blur together and before it becomes very difficult to write that 20th motherboard review and make it sound like something other than a total cut and paste job of previous articles.

The typing part takes me about 4 to 5 hours on a newer platform for a long winded article. As time goes by the time it takes to write them goes up and the amount of text goes down. You can only say certain things so many times. I try to approach each article as though each one could serve as a first article for anyone new to our coverage but that's harder than it sounds to actually accomplish.
 
Where are you getting this from??? They were touting RENDERING benchmarks and talk about going head to head with Broadwell-E....and they nailed that. I never got the impression this was going to go toe to toe with the fast quads. Where were you people?


argh?


right from the CEO's mouth.

websites have been saying it from months

http://www.digitaltrends.com/computing/amd-ryzen-cpu-leaked-roadmap-three-classes/
That said, AMD is targeting enthusiasts with the first wave of its Ryzen processors
 
Thanks for the review! I will be waiting until more data becomes available, especially to see what happens with the weird discrepancies seen. I am hopeful some of the bugs can still be worked out, leading to better performance.

I am disappointed in clockspeed, however. I was really hoping for 4.3-4.4ghz.
 
Well that is the problem isn't it, that is not the way AMD marketed this product, they marketed as gaming enthusiast, overclock enthusiast products. Neither are true at this point. Its still at Sandy bridge level of gaming performance a 4 core part that is 6 generations old!
Did they? The first benchmarks they showed us were of Blender. That's hardly a game. Actually AMD gave us benchmarks about every product out there that wasn't a game.
 
Yeah, and they will. Can you find me something that says it'll smash a 7700k in gaming? It's not a BAD gaming chip. But 90% of their marketing was competing against Broadwell-E.


It doesn't matter, that is not how they showed it, they showed it in game tests going toe for toe with Broadwell e and it doesn't, it can do it in specific games but gets trounced in others!

All Intel has to do is here, you want the best you still have to pay us more. Ok we will cut our prices down just a hair to keep status quo.
 
Did they? The first benchmarks they showed us were of Blender. That's hardly a game. Actually AMD gave us benchmarks about every product out there that wasn't a game.


Missed what Dr. Su stated who wants ryzen? I guess so, haven't been listening to their presentations with DeusX, BF1, Star Wars Battlefront and few others?

They have shown many games without anything outside of Broadwell - E close to or beating it, that did not happen either. So they cherry picked games and gave a false sense of confidence to potential consumers.
 
Missed what Dr. Su stated who wants ryzen? I guess so, haven't been listening to their presentations with DeusX, BF1, Star Wars Battlefront and few others?

I've heard a lot of pomp about gaming performance from VEGA....are you sure you're not mixing up your pressers?
 
I'm becoming a bit more convinced the gaming performance is partially due to (lack of) time spent at boost clock frequencies. Some slides I've been reading from AMD clearly show delineation for ">2 core boost" versus "2 core boost".

My experience with a 4C8T chip in gaming shows a lot of games "lightly" use a large number of cores, but rarely saturate them fully. I speculate this light activity "tricks" the CPU into avoiding boost mode. The intels, especially with the BIOS setting to use one multiplier for all cores with regards to boost, just tends to stay at the boost freq. At least in my experience.
Perhaps AMD is being very conservative with boost clocks to ensure stability with marginal cooling.

I'd love to see a clock freq and core usage vs. time while running games to see if this has merit.
 
Missed what Dr. Su stated who wants ryzen? I guess so, haven't been listening to their presentations with DeusX, BF1, Star Wars Battlefront and few others?

And note...I'm a little bummed it's not better in gaming, but I'm also not shocked, because I watched all their benchmark releases and it was painfully obvious they were targeting the workstation market with this release.
 
Back
Top