AMD Ryzen 1700X CPU Review @ [H]

cptnjarhead said:
I'm leaning more towards optimization... than the bottle neck theory..while these are hand picked... it does leave questions.
Click to expand...
By no means does it look like a actual CPU bottleneck of any sort, if anything it looks like Deus Ex ran slightly lower settings.
 
cptnjarhead said:
I'm leaning more towards optimization... than the bottle neck theory..while these are hand picked... it does leave questions.
Click to expand...

What you're seeing there, and what I've seen on a lot of Ryzen benchmarks, is that Ryzen actually performs a hair better under partially GPU limited scenarios. Fully GPU bound, and all scores are the same. Fully CPU bound, Intel stomps Ryzen. In between, sometimes Ryzen gets a little something extra, for whatever bizarre and probably random reason. It doesn't mean much of anything. In fully CPU-bound scenarios it still gets its *ss handed to it, and in fully GPU-bound scenarios, nobody gives a sh*t about the CPU anyway.
 
cptnjarhead said:
I'm leaning more towards optimization... than the bottle neck theory..while these are hand picked... it does leave questions.

r_600x450.png



r_600x450.png




r_600x450.png
Click to expand...

That's very interesting. Granted, I read the review, and Ryzen lost in all the other games. But the fact that it is winning is some is hope that the performance is there and maybe future patches can bring out all the potential.
 
cybereality said:
That's very interesting. Granted, I read the review, and Ryzen lost in all the other games. But the fact that it is winning is some is hope that the performance is there and maybe future patches can bring out all the potential.
Click to expand...

Without being a butthole (I hope ) Ryzen is what it is. The next rev will clear up the most obvious issues but it's not going be any software patch that does this. I accept and am happy with this Ryzen for what it is. But expecting more from software changes is in my opinion a fools waiting game.
 
Simplyfun said:
Without being a butthole (I hope ) Ryzen is what it is. The next rev will clear up the most obvious issues but it's not going be any software patch that does this. I accept and am happy with this Ryzen for what it is. But expecting more from software changes is in my opinion a fools waiting game.
Click to expand...
Hey now. Nobody from AMD ever claimed Ryzen was equal to Intel in gaming overall. They only said it was competitive. In ain the vast majority of games it us within 10% to 25% of Kaby Lake. That is competitive. Some fanboys here and elsewhere made hyped claims, but AMD did not. Ryzen is not a disappointment only fools and jackasses would say that. My only issue as an early adopter is the prmitive quality of the motherboard bioses.. I had two successful installations on two different bioses that were hosed because of instability of the bios. That was without any attempt at overclocking.
 
There are plenty of reviews showing two interesting "facts": Ryzen has lower framerates vs. the various intel chips at most resolutions, but most significantly at lowest resolutions (cpu bound); these same reviews show the intel chips at 98%+ cpu load while Ryzen is at 35-50% load per core.

It seems like Ryzen is leaving a lot of performance on the table. However, none of this makes a Ryzen build unplayable. It just opens up a line of questioning which should focus on WHY the Ryzen cores aren't being loaded to a higher level.

I'll be buying and making a Ryzen build...once the mobos become available. ;)
 
c3k said:
It just opens up a line of questioning which should focus on WHY the Ryzen cores aren't being loaded to a higher level.
Click to expand...
They are, simply Windows shuffling threads around and reported CPU load being average over time makes this misleading.
 
fuzzylogik said:
I was reminded that the fact that there's even a discussion of an AMD CPU VS an Intel CPU is a huge success on AMD's part. That hasn't been the case in a very long time!
Click to expand...

That's certainly true. Bulldozer was a giant steaming pile of sh*t. Ryzen is a huge improvement for AMD. I think that's getting lost in the noise right now. Yeah, Ryzen has a schizoid personality at the moment. It does great in some things and not so great in other things. But that's a huge step up from just plain sucking all across the board. There is actually a high-performance case for why you would want to buy AMD again. I'm very pleased with that. God knows Intel has needed a kick in the *ss for years.
 
DuronBurgerMan said:
What you're seeing there, and what I've seen on a lot of Ryzen benchmarks, is that Ryzen actually performs a hair better under partially GPU limited scenarios. Fully GPU bound, and all scores are the same. Fully CPU bound, Intel stomps Ryzen. In between, sometimes Ryzen gets a little something extra, for whatever bizarre and probably random reason. It doesn't mean much of anything. In fully CPU-bound scenarios it still gets its *ss handed to it, and in fully GPU-bound scenarios, nobody gives a sh*t about the CPU anyway.
Click to expand...

I will give you "bizarre" but not random. Something is not right... look at Crappy bulldozer, sucked all the way across the board no matter what resolution... - examples:
I'm just not buying into this bottleneck thing where Ryzen is concerned.


aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9NLzEvMzEwNTM3L29yaWdpbmFsL2YxLTIwMTEtMTY4MC5wbmc=


aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9NLzIvMzEwNTM4L29yaWdpbmFsL2YxLTIwMTEtMTkyMC5wbmc=


aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9NLzMvMzEwNTM5L29yaWdpbmFsL2YxLTIwMTEtMjU2MC5wbmc=
 
cptnjarhead said:
I will give you "bizarre" but not random. Something is not right... look at Crappy bulldozer, sucked all the way across the board no matter what resolution... - examples:
I'm just not buying into this bottleneck thing where Ryzen is concerned.
Click to expand...

That is pretty much what I have tried to mention in any post here or outside. It's obvious that the "problem" is more of a throughput issue and not a processing issue. Latency and cross talk drags the cores down when all they are doing is simple page refreshes as quickly as possible. But the platform itself has legs and in the "Bottle necked" bench marks for whatever reason it tends to (though not always) have the better numbers. The difference is much smaller, but it was also at those settings I think the difference matters most. If your missing 20FPS from 180 down to 160 you aren't going to miss it. If your missing 3 frames at 60, you probably still aren't going to miss it but, but if you did miss one of the 2 it would be those 3 FPS.
 
It seems to me AMD did a really good job overall and regardless of the reason for some of the inconsistencies you wouldn't be making a dumb decision to buy Ryzen, especially given the price/performance ratio. I rarely buy first generation products because of stability/quirk concerns, but once AMD releases a rev with a bit more polish and some more mobo options I very well might get a setup. I dont game nearly as much as I used too and do alot more workstation-like work, which Ryzen is perfect for. More cores for less cheese. The only thing Im a bit concerned about it is virtualization which may need some more time for AMD to work out the current issues.
 
Somebody was saying "have fun with your spreadsheets" to the folks buying Ryzen in another thread. It's not like that at all. Sure, a purist gamer gives no f*cks about encoding performance, or rendering, or any other sh*t other than pure FPS, cranked to the f*cking max.

But a lot of folks, myself included, do other sh*t with our machines. And not spreadsheets. For instance, I am a UI Developer & Graphic Designer, and also a trance DJ/Producer (no, not a bedroom DJ - a f*cking club DJ).

Now for UI dev, I do a lot of Angular work, which these days is done in Typescript. Sh*t gets compiled down to JS. I think it's kind of stupid, actually, but whatever. Point is, it has to compile while I'm running a local server, with Eclipse serving up a restful service, and like 4 or 5 other applications open for editing sh*t. I've got threads out the ass running at any given time, and THEN I'm asking for the machine to compile something on the fly whenever I save a change. I'm going to be dollars to donuts Ryzen 7 will trounce a 7700k in that scenario, and not even by just a little bit.

Then later I fire up FL Studio to so some music production, right? I've got anywhere from 16 to 30 channels in an average song. Each channel has a VST, or a MIDI in/out running to a hardware synth. VSTs stacked on top of each other? Huge multi-threaded workload. Advantage: Ryzen. Again, not even by a little bit.

I do some video blogging with a green screen (it's political sh*t so I'm not posting the link here). Requires some After Effects work, then rendering & encoding. Prime multi-threaded workload. Sometimes these renders take f*cking forever. Ryzen gives me tangible immediate benefits over quads.

Now then somebody says... "well hey BurgerMan, you gonna lose 15 fps in Tomb Raider" like I give a sh*t. I bought a 1080 ti, I ain't hurting anyway. I don't care. Only thing that worried me was that gaming performance could be an indicator of major future problems with Ryzen, but this doesn't seem to be the case. It appears to be a case of architectural quirks (some of which may be mitigated in the software in the future) and just what happens when you've got a battle of more cores vs. higher clocks. We went down the same road with quads vs. duals once upon a time.

7700k ain't a bad chip. If you're a gamer first, Kaby Lake is your sh*t. If you're doing anything like what I'm doing, either go Ryzen, or let Intel bend you over for their 8 core and 10 core parts (which are great, but overpriced). If you're a spreadsheet guy... well, you sure as hell don't come to [H]ard|OCP anyway.
 
Why not clock the intel chips at the same speed? Is this truly a fair comparison with a 1.0ghz advantage? Just asking.
 
Konceptz said:
Why not clock the intel chips at the same speed? Is this truly a fair comparison with a 1.0ghz advantage? Just asking.
Click to expand...

That's like asking to amputate a leg off of every participant in a race because someone with an artificial leg wants to compete 'fairly'.

Disclaimer: I have an artificial leg and if you're a special snowflake who is offended by my post just keep it to yourself. :whistle:
 
Algrim said:
That's like asking to amputate a leg off of every participant in a race because someone with an artificial leg wants to compete 'fairly'.

Disclaimer: I have an artificial leg and if you're a special snowflake who is offended by my post just keep it to yourself. :whistle:
Click to expand...

Do you keep your coke stash in a secret compartment in there? I would, if I liked coke.

I agree, test them at stock and test them as they are typically installed (OC).
 
Explain to me how clocking all the CPUs @ say 3.8ghz is bias??
Click to expand...

For the same reason disabling 4 cores on the Ryzen 7 CPUs when compared against an i7 7700k would be bias.
 
I think there are two comparisons of CPU's that are generally worth while with some other configurations being interesting in an academic sort of way. CPU's should be run at their default settings with the rest of the system's being as similar as is possible This will tell you which is better in a given application the way most people would use them. For the enthusiast, only the overclocked values they can expect really matter. So you take your best or average achievable clock speeds for an AIO or custom loop system and compare the two with everything else being as close as possible. For the sake of argument, here's what Ryzen should be pitted against:

Core I 7 7700K@Stock vs. Ryzen 1x00 Series @ Stock
Core i7 7700K@ 4.8GHz vs. Ryzen 1x00 Series @4.0GHz.

Examples of Ryzen at 4.1GHz or the Core i7 7700K @ 5.1GHz are also useful in a sense but those are beyond the speeds most people will ever achieve. Given the core count, comparisons against the Core i7 6900K @ stock and 4.3GHz are valid. That's muddier waters of course given the disparity in the feature sets between platforms. Results of Ryzen and Kaby Lake are identical clock speeds are academically interesting as it highlights the IPC as the major variable between the two of them. This information is misleading as these CPU's aren't likely to be used at those frequencies very often if ever. The value in doing so is that it establishes a baseline as a predictor of performance at different speeds. If one CPU has a slight advantage over the other, you can determine what speeds you need to achieve for one processor to overtake the other.

The Pentium 4 sucked. However, if I had one at 8GHz it would have smoked any Athlon 64 that wasn't remotely in the same ballpark in terms of clock speed. However, LN2 results at 5.0GHz showed that it still got trounced in most tests by substantially lower clocked Athlon 64's if I recall correctly. This is the type of thing IPC measurements can tell us. If you see a 10% reduction in IPC in one architecture over another in a given application or application set, then you can predict what models or what clock speeds you need to hit when buying a CPU with overclocking in mind. This isn't an absolute way to calculate performance. Checking into actual test results is the best thing to do when shopping for CPUs. You also want to get a bead on the average overclocking range of these chips to figure out if you can meet your goals. Knowing the IPC numbers as a baseline helps narrow that search or stand in when you don't have exact clock ranges represented in the various reviews which does happen sometimes.

If a CPU architecture is behind another one by 10% but overclocks 30% higher it's easy to figure out which is the faster option. Even if the stock clocks are a given amount lower than the slightly faster CPU. So there is value to base IPC numbers.

Similarly, people bitch about CPU limited gaming benchmarks not being representational of the performance you would get in actual games. When people say that they are being myopic and only seeing the CPU through one particular filter. While they are correct in their line of thinking, they are only correct to a point. This is a case where the numbers you get at face value aren't telling you the whole truth. Knowing which CPU is faster is important when you know very well that you could end up in a situation where the limitations of the CPU become apparent. Interestingly, these happen only at the opposite ends of the spectrum. At extremely low resolution settings or at very high resolution. People don't really make the connection as to why this happens or what it means for them.

If I game at 640x480, it doesn't really matter if my Ryzen 1700 can only achieve 342FPS in Doom while my Core i7 5960X can hit 430FPS all day long. No monitor refreshes that fast. It is also doubtful that either CPU would allow a large enough frame rate reduction in any part of the game for me to perceive it. Therefore, saying that these resolutions do not matter is somewhat accurate. It's certainly accurate if gaming at those low resolution settings or quality levels. On the other hand, knowing which CPU is faster at the task when CPU limited helps us at the top end of the spectrum. Here's what people don't realize: Let's say I'm gaming at 7680x1600. Let's say I've got two systems to choose from to do this and that budget isn't a factor in making my decision on which box to keep. If both systems are rocking 1080Ti SLI, then the only variables that remain are the CPU and anything else unique to their particular platforms. At these and other extremely high resolutions, the CPU may again make a difference because its the only hardware variable that differentiates them. Not every game ever tested will show a benefit to the CPU that's faster in CPU limited scenarios because the game's frame rate might be capped or limited by other factors. IF however, the CPU does make a difference (and it can) then I know which CPU or system I should choose.

Academically speaking, this shit isn't that hard to figure out and I don't understand the sheer volume of discussion beating these topics to death. The fact is that most of the time, it makes little difference what CPU you have in your machine. The reviews exist to provide information for the reader to draw their own conclusions based on the applications and games they use. To that end, I don't think it makes much of a difference that Ryzen is generally slower in gaming than Intel's CPUs are. The cases where it would matter are few and far between. However, if you think you are going to hit those limits then your choice becomes clear. There are two things after all is said and done that really matter without getting into brand loyalty and bullshit like that. 1.) Is the performance difference worth the added cost to go to the better performing product? That question is answered by determining how often your usage will put you into a situation to benefit from the more expensive product. 2.) In trying to predict the future, which architecture is better for the long haul. That's the harder question to answer because we can't see the future. All we can do is take the evidence we have now and speculate as to how it will shape the future. On that point, people really should be honest with themselves.

We know that Bulldozer is a better gaming CPU today than it was when it launched. How do we know this? Because it's doing better in benchmarks with any game that makes decent use of any CPU beyond the standard 4c/4t configuration. 4c/8t Core i7's only work better in those situations because of the MASSIVE IPC difference between Bulldozer and Skylake / Kaby Lake. So if your choice is between a 4c/8t Intel CPU and am 8c/16t Ryzen the real question becomes: How long will you keep the CPU? If you upgrade often the 7700K might be better today. If you keep a CPU for 3-5 years or more then you might really want to go with Ryzen.
 
Dan_D said:
So if your choice is between a 4c/8t Intel CPU and am 8c/16t Ryzen the real question becomes: How long will you keep the CPU? If you upgrade often the 7700K might be better today. If you keep a CPU for 3-5 years or more then you might really want to go with Ryzen.
Click to expand...

QFT. Thank you for bringing sanitary to the nonsensical discussions and trolls that have spawned countless threads...
 
Dan_D said:
... Academically speaking, this shit isn't that hard to figure out and I don't understand the sheer volume of discussion beating these topics to death. The fact is that most of the time, it makes little difference what [product] you have in your [inventory] ...
Click to expand...

QFT, this applies to so many things in general.
 
Thanks Dan_D , that all makes a lot of sense.

One point I think that is missed, when gaming it's really the whole system that affects performance. Though there can be one part that is limiting, sure, every single component can make a difference.

Case in point, I run on a 4K TV and even with a Titan XP I felt that the machine wasn't optimal. I ended up swapping out the i5 to an i7 and saw a small but measurable difference in performance on many games (about 10%). So even on max settings with 4K resolution, it's not only the GPU doing work.

Or in the case of Surround/Eyefinity, the GPU is seriously stressed but so is the CPU. With that higher FOV (and, to a lesser extent, the same thing can happen on ultra-wides) you are rendering more objects, so more draw calls and processing on the CPU. It all contributes.
 
You must log in or register to reply here.
Back
Top