Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

[Insert_Nickname]

I'm a developer who has no overhead room for issues like this in my dev environment or playing kernel lottery games whereby I am restricted by other software packages :
https://community.amd.com/thread/215773?start=120&tstart=0

I hope this doesn't come of as too blunt, but if you can't afford issues, then why are you looking at this in the first place? Much better to stick with proven and validated HW for the time being. Give Ryzen a year or two to work out the kinks before considering.

Yeah, final validation being known in business as a process that usually consists of months if not years of testing, debugging, and rounds of fixes from the manufacturer before something is ever put into actual production environments. Anyone who actually works in tech or enterprise would know of this process and understand what validation truly means for new (professional) equipment.

Various business entities have different levels of resources to dedicate to such true validation tasks thus why proven companies and products capture business instead. But i guess (((validation))) means something else to people outside these environments. I guess that just means (Pro) slapped in front of a micro-architecture. Cute bants. However, it's not what I came here for.

If validation is that important to you, then you have no business looking at consumer grade hardware or building yourself. A proper, certified workstation from a reputable manufacturer is what you need.

 

[Topweasel]

So, AMD is going to allow an OEM to launch their groundbreaking HEDT processor before they officially detail it themselves? Sounds like an epic fail to me both from an OEM standpoint and for AMD. Pre-order Launch for a product that isn't even detailed or officially released through an OEM.. Is this what they did with Vega through Apple's exclusive licensing too? The difference being that Alienware is a low yield enthusiast platform vs. a mainstream product like Apple. A good way to put a sour taste in the community that is likely to buy your product.

Look yield to Dell, maybe everything that they can produce for AMD. Take X299 vs i3-i7. In a given month Intel may ship 10 million of the later and only 100k of the former. AMD shipped 1-2 million R7's in march when demand was at its highest. 1% of that is 10-20k which might be the size of a single volume order from Dell. Dell is the perfect launching partner. They have the support of a large OEM, makes money off the exclusivity deal and basically has Dell offering to take the whole first shipment. That is a huge win for them and takes a lot of the risk of the productline off their shoulders. Also we are talking about a week or 2 difference for a product that we didn't even really know existed till early last month.

I will be watching this closely as this will keep my highly cautioned from considerations of plunking down cash for such a pricey platform. If that's the way they go about things, I'd much rather spend pennies on testing out their ryzen platform while the HEDT gets the kinks worked out of it across a year or two, and at a later time spend dollars.

As you have been responded to before by others. It's disingenuous to talk about kinks in platforms considering A.) It's running the same chipset as the x370 boards. B.) Litterally the only problem that has ever "plagued" the Ryzen platform has been memory speeds above Jedec, in which many people didn't have any problems at all and by now almost everyone can run their stuff upwards of 3200+ without issues and 4k with a little effort. Same thing applies on any platform. This isn't a bug filled mess.

They should know better about making official, timely, and full launches based on how this Vega foolishness has gone down.

I mean you can't hurry these things. AMD is releasing FE for professionals to basically show how close they finally are and it,s probably going down worse than if they stated it was delayed for one last quarter. and as I stated above. We have known about Vega and it's already 6 months or so late. We weren't even sure TR existed till mid May and AMD hasn't even given a real launch date. Its to completely different environments.

 

[ub4ty]

Look yield to Dell, maybe everything that they can produce for AMD. Take X299 vs i3-i7. In a given month Intel may ship 10 million of the later and only 100k of the former. AMD shipped 1-2 million R7's in march when demand was at its highest. 1% of that is 10-20k which might be the size of a single volume order from Dell. Dell is the perfect launching partner. They have the support of a large OEM, makes money off the exclusivity deal and basically has Dell offering to take the whole first shipment. That is a huge win for them and takes a lot of the risk of the productline off their shoulders. Also we are talking about a week or 2 difference for a product that we didn't even really know existed till early last month.

I feared this was the case.
I guess I'll just stay away from coverage until they actually give a consumer release date and the platform/mobos are extensively tested/benchmarks.
Thank you for a solid detailing/response on this issue.

 

[ub4ty]

I hope this doesn't come of as too blunt, but if you can't afford issues, then why are you looking at this in the first place?

It doesn't come off as blunt. The mention of validation without seemingly understanding what that means in enterprise and now the mention of affordability come off as what they are : snide remarks. Enterprise customers validate a manufacturer's new products in-house even after a manufacturer (claims) it's been validated. Most commonly, there are tons of show stopper bugs that prevent it from being deployed in a runtime environment. This validation can take months if not years. This happens for equipment well in excess of millions of dollars all the way down to a $0.10 asic. It's exactly what happened pre/post Epyc release and its a (professional) product. Sometimes products pass through in weeks with minor addressments. sometimes months. Sometimes years. It all depends on the number of bugs found, how severe they are and how long it takes the manufacturer to fix them. Consumers are getting more of a taste of this frustrating process as hardware is more rapidly pushed out the door. You're now validation testers upon release. It wasn't always this way. Consumers used to hold manufacturers to higher standards. Now, they openly defend and debate anyone critical of such botched releases as several of you are doing here now. As a result, you get to enjoy the lovely and frustrating inhouse validation process of their releases.

Being that I've actually sat on releases of such (pro)/(enterprise) products, I know exactly the kind of crap that can get rubber stamped with that label and be riddled with quite serious bugs. Affordability has nothing to do with what is being discussed. So, I take it as a snide jab to try to strengthen an incorrect position and commentary and it failed.

Much better to stick with proven and validated HW for the time being. Give Ryzen a year or two to work out the kinks before considering.

I recall this is exactly what I stated. Sorry for speaking my mind, I thought this was a forum.

If validation is that important to you, then you have no business looking at consumer grade hardware or building yourself. A proper, certified workstation from a reputable manufacturer is what you need.

What's important to me in my at-home development environment is different than what's important to me in my professional capacity.
Tons of developers have clear reasons for consumer grade hardware/building systems themselves as do tons of businesses.
Certified workstations, if you've ever actually administered them equals high cost service contracts and low turn around time on repairs.
How low a turn around time equals how much you're willing to pay for tiered service contracts. It's not a panacea nor is the hardware/software bullet proof (especially the new h.w/s.w). You're buying a service contract, not somehow getting gold plated hardware.

It's a forum. Were talking about CPUs. Developers and various businesses can and do have every right finding an equilibrium between cost/function/features/support when it is known how exactly pro certification/service contracts work. The real-world of enterprise isn't some idealistic rubber stamped panacea. It's littered with the same crap you see on the consumer side.

Every year the quality of releases drifts further and further south. You come on places like this and there's no wonder why

You've managed to somehow make snide remarks about my commentary yet ultimately conclude the same thing I have stated.
Congratulations.

 

[Veradun]

...

What's the problem? You can buy an i7 with the same core count at triple the price and disable HT because bugs and be well with your coding. No?

 

[Atari2600]

You've managed to somehow make snide remarks about my commentary yet ultimately conclude the same thing I have stated.
Congratulations.

You are dumping on something that is <6 months old for not being mature.

Any wonder people are mystified.

 

[Atari2600]

Or are Intel up to their old shenanigans?

https://www.pcspecialist.co.uk/forums/showthread.php?53124-AMD-ryzen/page7

I now see PC Specialist have the new Intels on sale... with memory speeds up to 3600 MHz. Completely beyond JEDEC spec and there cannot be much evidence of long term stability. So any excuses for only low memory speeds available for Ryzen go out the window.



I think its fairly safe to say they've been paid off and Intel are indeed up to their old tricks. For shame. Hopefully someone from either PCSpecialist can and resolve this, or someone from AMD sees this and initiates the appropriate legal action.

 

[cmvrgr]

Intel fans should say thanks and they have to be happy and grateful to AMD because AMD is the reason that they will get cheaper and greater performance from intel too. So negativity about ryzen is not necessary.

 

[krumme]

Rumor
http://wccftech.com/amd-takes-10-4-...17-largest-single-quarter-share-gain-history/
"AMD Takes 10.4% CPU Share From Intel In Q2 2017 In Passmark Survey, Its Largest Quarterly Share Gain In History"

 

[VirtualLarry]

Rumor
http://wccftech.com/amd-takes-10-4-...17-largest-single-quarter-share-gain-history/
"AMD Takes 10.4% CPU Share From Intel In Q2 2017 In Passmark Survey, Its Largest Quarterly Share Gain In History"

Wow, pretty awesome if true. My personal opinion? Ryzen rocks! It's a great daily-driver CPU, plenty of multi-tasking power, and not too bad single-threaded either, though not always as fast as overclocked (higher frequency, mostly) Intel CPUs.

 

[moinmoin]

Keep in mind that that's just the Passmark stat of benchmarks submitted to them, not the actual market share.
https://www.cpubenchmark.net/market_share.html

 

[Veradun]

Keep in mind that that's just the Passmark stat of benchmarks submitted to them, not the actual market share.
https://www.cpubenchmark.net/market_share.html

Exactly as Steam Hardware Survey it offers some kind of gross estimation of market trends and you shouldn't use it strictly. Qualitative, not quantitative.

 

[Lodix]

Samsung has disclosed that their next 14nmLPU provides 15% better performance, which is probably the same 14nm+ AMD will use from GF in their next zen core revision.

 

[tamz_msc]

Okay so I've been reading about the SPEC benchmark suite, and the more I read it, the more I wish it were a part of reviews featuring consumer CPUs, especially the updated 2017 suite. The obvious problem with it is compiler optimizations, and there are benchmarks within it, like integer libquantum, that have been heavily gamed by the Intel Compiler. Still I wanted to see how Ryzen fares with ICC17, and so I went ahead and bought this article from heise.de, with money I never got to spend on the recently concluded Steam Summer Sale. They've actually run the SPEC2006 benchmark suite on the Ryzen 7 1800X, i7 6900K and i7 7700K. And the results are quite interesting.

Now the integer scores are heavily skewed by libquantum, even so with ICC2017 and AVX2(!), the 6900K is only about 16% faster than the 1800X in SPECInt-2006 single-threaded. But the tides completely change in the SPECfp-2006 single-threaded, and the 1800X is now 12% faster than the 6900K! Though the multi-threaded scores are consistently around 25-30% better on the 6900K.

Now I wish upcoming reviews of Threadripper and the high core count Skylake-X chips, especially those of AnandTech, included the newly released SPEC2017 suite.

 

[moinmoin]

Exactly as Steam Hardware Survey it offers some kind of gross estimation of market trends and you shouldn't use it strictly. Qualitative, not quantitative.

No, it's worse than the Steam Hardware Survey. With Steam it's just a subset, but at least that hardware is definitely there. With Passmark every benchmark submission is counted, even if there are multiple ones by the same person using the same hardware. AMD is most certainly increasing market share, but the nature of that Passmark stat won't help guessing either quantity or quality of that increase, just how often AMD chips were benchmarked relatively to Intel chips.

 

[Veradun]

No, it's worse than the Steam Hardware Survey. With Steam it's just a subset, but at least that hardware is definitely there. With Passmark every benchmark submission is counted, even if there are multiple ones by the same person using the same hardware. AMD is most certainly increasing market share, but the nature of that Passmark stat won't help guessing either quantity or quality of that increase, just how often AMD chips were benchmarked relatively to Intel chips.

I don't get it. Do you think people suddenly started submitting more if owning AMD chips or stopped submitting if owning Intel chips?

 

[Mopetar]

I don't get it. Do you think people suddenly started submitting more if owning AMD chips or stopped submitting if owning Intel chips?

You don't think a new CPU (and architecture) would see more benchmarks than usual, especially as most people buying it are enthusiasts and will play around with it more?

AMD is going to gain market share for sure, but as pointed out this is a distorted metric by which to gauge that gain.

 

[Veradun]

You don't think a new CPU (and architecture) would see more benchmarks than usual, especially as most people buying it are enthusiasts and will play around with it more?

I don't know. Has Kaby Lake told us something on this regard? What about Skylake? Go on back wherever you want, there is a never experienced before surge on the red side, and that means we get a qualitative indication of a new market trend as I said earlier.

 

[Topweasel]

I don't know. Has Kaby Lake told us something on this regard? What about Skylake? Go on back wherever you want, there is a never experienced before surge on the red side, and that means we get a qualitative indication of a new market trend as I said earlier.

I couldn't tell you there but what I can tell you is that fresh install of Steam on my system hasn't resulted in a Steam Survey request. There are so many variables that go into the steam survey including voluntarily running the survey (which I am personally against doing) that it isn't a great tool. It is a tool though.

I would add that you also have to look at the typical purchasers and users. I would expect more 7700k users to install steam than 7700 users. The 7700k is solely target at enthusiast and therefore purchasers of that CPU are more than likely PC gamers and therefore are very likely to have Steam installed. So in terms of units sold vs. systems on steam I expect the 7700k ratio to be very high. Whereas the 1700 and higher have had something we haven't seen which is competitive 8 core CPU's on a consumer priced platform. A low cost 8 core CPU solution has a market outside just gaming. I myself purchased mine for mostly VM work, though I game, I am not seeking the absolute best gaming performance for the money. But chances are there are a higher percentage of users purchasing a 1700 or higher that are not gaming and haven't installed Steam than 7700k users. The 1500 through 1600x probably have higher ratios than the 1700+.

There is a third aspect. OEM vs. Homebuilt. As it is right now there isn't really a option or at least many options for OEM systems which accounts for a ridiculously large percentage of systems purchased and even in Steam. Sure the ratio is going to be better in Steam but still almost insignificantly higher. So even if the the enthusiasts purchases started a major shift into Ryzen's favor it wouldn't be hard for a small spike in OEM system purchases to mask any growth from homebuilt systems.

TLDR; New systems don't mean new surveys, Ryzen's strength will equal smaller ratio of systems vs systems with steam, and OEM sales are so dominant small shifts on OEM sales can mask any enthusiasts traction in survey's.

 

[Veradun]

I couldn't tell you there but what I can tell you is that fresh install of Steam on my system hasn't resulted in a Steam Survey request. There are so many variables that go into the steam survey including voluntarily running the survey (which I am personally against doing) that it isn't a great tool. It is a tool though.

I agree with that, Steam HW survey is a joke, i've never got a survey in like 7 rigs :>

Passmark is just a bit better, as it at least is representative of its own userbase unlike Steam

 

[richierich1212]

My Ryzen system hasn't gotten a steam hardware survey request yet my Dell Laptop with i5-6300u does. Go figure. ‍♂️

 

[french toast]

Right so I have just read this 30 game benchmark from techspot, comparing 215$ r5 1600 vs 415$ i7 7800x.
Note that the Intel is a hedt CPU with quad channel memory and uses a whopping 380$ custom loop cooler vs the half price 1600 which uses dual channel memory and a free cooler with the CPU...Oh and the motherboard is significantly cheaper with AM4 vs x299.
https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

At stock speeds the i7 is a bit faster due to the obvious higher clocks, but once both are OC to 4ghz and 4.7ghz respectively both have the same gaming performance! Also the ryzen offers significantly lower power consumption and so much better perf/W whilst destroying it in perf/$.
Actually when overclocked 1600 is not much behind the fastest gaming CPU 7700k! (9℅@4.9ghz) That's pretty incredible if you ask me and not something I expected having not checked up on ryzen for a couple of months.

"Once overclocked, the $215 R5 1600 was just 9% slower than the 7700K, while it matched the $415 7800X."
"Ryzen will hit 4GHz with the box cooler but it will be a more mild experience with a $20 aftermarket cooler like the Cooler Master 212, so keep that in mind. The 7800X on the other hand cannot be overclocked to 4.7GHz using a 240mm AIO closed loop solution. Instead, it required a $380 custom loop setup to achieve that result"

Edit; corrected dual channel for single for ryzen (mistake).

 

[DrMrLordX]

Is that 1600 using single-channel or dual-channel RAM?

 

[french toast]

Is that 1600 using single-channel or dual-channel RAM?

He seems to be using 32gb ddr4 3200 for skylake x and half that for ryzen, I wish it was more specific but seems to be 8gb sticks? For quad channel on skylake x and dual on ryzen, although he mentions he didn't have 32gb stick for ryzen which was odd.

Edit; correction he said 32gb kit not stick.

 

[moonbogg]

Right so I have just read this 30 game benchmark from techspot, comparing 215$ r5 1600 vs 415$ i7 7800x.
Note that the Intel is a hedt CPU with quad channel memory and uses a whopping 380$ custom loop cooler vs the half price 1600 which uses dual channel memory and a free cooler with the CPU...Oh and the motherboard is significantly cheaper with AM4 vs x299.
https://www.techspot.com/review/1450-core-i7-vs-ryzen-5-hexa-core/page9.html

At stock speeds the i7 is a bit faster due to the obvious higher clocks, but once both are OC to 4ghz and 4.7ghz respectively both have the same gaming performance! Also the ryzen offers significantly lower power consumption and so much better perf/W whilst destroying it in perf/$.
Actually when overclocked 1600 is not much behind the fastest gaming CPU 7700k! (9℅@4.9ghz) That's pretty incredible if you ask me and not something I expected having not checked up on ryzen for a couple of months.

"Once overclocked, the $215 R5 1600 was just 9% slower than the 7700K, while it matched the $415 7800X."
"Ryzen will hit 4GHz with the box cooler but it will be a more mild experience with a $20 aftermarket cooler like the Cooler Master 212, so keep that in mind. The 7800X on the other hand cannot be overclocked to 4.7GHz using a 240mm AIO closed loop solution. Instead, it required a $380 custom loop setup to achieve that result"

Edit; corrected dual channel for single for ryzen (mistake).

https://www.youtube.com/watch?v=UfNMn7RWgLw

There's the video link. I think that 1600 sort of came in under the radar really. IMO that is the CPU to get for almost everyone. I can't believe how well it does. I really didn't expect this. The gaming performance of Ryzen is not compromised in any meaningful way at all, especially when comparing to Intel's botched X299 chips. $200 bucks for that 1600. I really hope this chip gets used in millions of gaming rigs everywhere and becomes totally famous for being the greatest chip of this new golden era. Mind blowing really. The 1600 basically destroys the entire X299 platform for gamers who refuse to remain stuck in the mud with 4 cores. It delivers the same performance at half the price and also makes coffee lake pretty much not worth waiting for since that will just cost more anyways and be about 10% faster for gaming. CRAZY!