Editorial AMD's Ryzen Debut: Onwards to the HEDT Market or The Stumbling Hype Train

[doxology]

So people are crying and raging because 1800X cannot beat 7700k in Gaming? what ever happen to people two years ago expecting Zen to be on par with Haswell or Slower and calling it a massive win? so we know know is as fast as Skylake IPC Clock for Clock... we also knew and expected 8 cores 16 Threads to not clock with high frequencies, so why all the hate while the 7700k that has 500MHZ more of Frequency and beating the 1800X in Gaming? i think for the price of current R7 lineup is well worth it and competes extremely well even in gaming with Intel's 8 Cores 16 Threads CPU's.

I for once will not be upgrading to Ryzen because my i5 6500 is more than enough for Gaming, besides games do not use more than 6 Threads and this will be the norm in the following two years. The real hype train in gaming in my opinion are the R3 and R5, but.... the frequency that AMD announced does seem quite lacking, In any case 8 Cores 16 Threads CPU's are aimed more at productivity than Gaming. I think R7 Lineup is quite Good since i always expected Zen IPC would match Haswell, and i knew very well that 8 Cores 16 Threads would have lower Clocks to keep TDP in Check. It was expected! The good news is that it forced Intel to lower their prices, which was a most needed thing.

If AMD can do R5 with 6/8 Cores without SMT i'm Gold for Ryzen. As gamer i got little interest 16 threads + extra price, besides games will still be using 4-6 cores in the next two years. But AMD really really need to raise the Frequency for R3/R5 or it would look very underperforming in gaming compared to what Intel has to offer at this moment, let alone what ever comes up from Intel. even if Intel CPUs cost more people will choose them just because is a tad bit faster. It is what it is.

In any case, no wonder why TPU has not been getting Rewiew hardware from AMD... looking at TPU past reviews lol.

Yeah noticed the same thing. They keep saying the Single Core Performance isn't as good and it is pretty clear that the base 7700k is clocked faster at default let alone turbo than the 1800x and that some sites are even using a 7700k clocked at 4.9GHz to show it is faster. They use words to tell us one thing but if you look at their slides we get a slightly different story. I want to see a 7700k clocked at 4ghz and an 1800x clocked at 4ghz and see how things actually play out clock for clock.

 

[Steevo]

Its slightly worse than I actually expected, the hype train isn't derailed, but it sure slowed down.

Just as I predicted the memory controller needs work, AMD should have gotten some great overclockers and some people who are good with BIOS's and put them together in a room with red bull or coke instead of some of their other events.

Its also VERY important to remember how fragile the IMC was for Intel on a few chips, and how shitty the original Hyperthreading really performed in all but the most cherry picked scenarios.

Dave, can we get some side by side X87 (SuperPI, Prime95) single and multiple cores, math run? How about some USB and SSD tests? Any deviation when a benchmark is run multiple times (does the AI learn anything, or does it need a kick from the highest prioity thread?) Does changing thread priority do anything?

 

[FR@NK]

There goes my dream of getting a used 6950x for cheap...

 

[cadaveca]

A lot of the other reviews I have seen have been similar with memory clocked at 2666. Not sure if it is the boards they got or what. Linus used an Asus ROG board I believe.

No, it's a lack of proper hardware. Reading a board's manual tells you how to get higher, if you have a bit of "high-level" OC experience. Linus's lack of explaining what is going on shows how inept he is. He failed both his users and AMD in that.

ASUS ROG board hit 3600 MHz memory, and there are screenshots here in another thread. It's not the board.


BTW, thew board that I posted a screenshot of is only rated for 2400 MHz with 2 DIMMs, and 2666 with only one DIMM. Yet I got 3200 MHz with tight timings with two sticks. No tricks needed... I simply enabled the XMP profile.

Being able to read a manual and properly select supporting review hardware should be very basic for someone trying to review hardware accurately for their readers/viewers.

Higher memory clocks I think should help in benchmarks.

Well, I got higher memory clocks, so I can tell you that it doesn't much. The CPU's cache is a bit of a bottleneck under SMT. Single-threaded workloads have enough bandwidth, only latency improvements are to be had.

One thing i have noticed from those that have gotten all three chips is that most of those sites got at least 100hz more from the 1700 than they could from the 1700x which seems kind of interesting.

My 1700X pulls 100W. If a 1700 pulls 35W less, as rated, then it makes sense, and isn't really interesting at all. Unfortunately, the included cooler won't be enough to push good clocks since at the top end we I am seeing 200W+ when overclocked. Again, this is pretty basic stuff as an oc'er.

Be interesting to see how it plays out once we have a platform, games, and apps all optimized with a mature platform I expect things could look much better than they do today on launch day.

Meh, I have seen enough already. Really though.

The challenge of pushing memory clocks will keep me interested for a little bit, but that's all.

There goes my dream of getting a used 6950x for cheap...

ROFL.

 

[R-T-B]

I don't know of anyone with half a brain in their head, that expected single core performance to match up with the consumer grade i7s. To compete in single core performance with other workstation CPUs? Absolutely, and they do exactly that.

Again...PROFOUND STUPIDITY....

Since when has the hype train had a brain?

OR did TPU editors got some special emails from Intel? The plot thickens....

Heh, no. We aren't that special. That would be cool though, if only to tell them where to shove it.

 

[Blueberries]

Single-threaded en/decoding, de/compression, pi, and PassMark when you do get samples please. Comparing a 6900k to an 1800x in Ashes of Singularity doesn't help anyone.

 

[Prima.Vera]

Very disappointing 1080p Gaming Performance

yeah, but on higher res is on pair with any CPU's out there. That being said, at the moment , for me, there is ZERO (0) reason to upgrade from my 3770K CPU, specially since I'm gaming on 3440x1440 res.
Oh well, will wait for another 5 years to buy a worthy CPU.

 

[goldfries]

That's the one. Figured there was something wrong. I want to see some premiere 4k exports.

Hi,

VIdeos were set to 1 minute.

 

[birdie]

A derailed Ryzen hype train has brought us these strange things:

• For best performance you suddenly need BIOS updates, games rewrites, etc. etc. etc. Just like with RX 480 which needed tons of work to show its true performance while so far the truth has been that NVIDIA destroys AMD in D3D 11 and earlier titles and AMD wins in some D3D 12 titles by the virtue of certain games being optimized and written for the GCN arch from the get go (read Doom). But even this advantage will soon be partially negated by D3D 12 ready NVIDIA drivers which will substantially improve NVIDIA's performance.
  
• TDP is enormous and even more so when being overlocked. It's nowhere near the advertised 95W. It's still miles better than with previous AMD architectures but the hype train certainly derailed here.
  
• AVX2 performance is halfed thus Ryzen 1800X is practically the same as 7700K in AVX2 workloads (x.265 encoding) even though it has two times more cores and substantially higher TDP.
• I wonder why I've never seen reviews prior to the launch of Ryzen which mentioned micro-stuttering in certain games with the Intel 7700K CPU. Suddenly Ryzen gives better fluidity while it's never been the case before. I call this BS.
• AMD has suddenly decided that most gamers play at 4K. Nope, it's less than 5%.

I was about to dive in but all things considered I will not be upgrading from my Intel Core 2500. Ryzen is an excellent workhorse for certain enterprise workloads but I have a simple desktop under the table. I will wait for Ryzen 2.0 with

• Better overall performance
  
• Reworked AVX 2.0 support (not sure it's possible)
  
• Better clocking out of the box
• Decreased TDP
• Largely improved IMC

or Coffee Lake which will offer six cores on the desktop.

Ryzen launch was certainly rushed but I do understand why AMD did that.

 

[Caring1]

The only Hype train to crash was that driven by gamers with their lofty expectations, the rest of us are happily motoring along on our AMD Bullet train to productivity.

 

[deemon]

is the intel compiler thing finally fixed? did you patch all the games tested prior to get rid of intel's intel-only CPU optimizations?
http://www.anandtech.com/show/3839/intel-settles-with-the-ftc

http://www.agner.org/optimize/blog/read.php?i=49#794

when it comes to intel, I never trust them... ever again.

 

[librin.so.1]

I am getting my [male genitatalia] hard over the compile times I'm going to get with this baby
*gagrles in ecstasy*

 

[Kyuuba]

Watch from minute 39:18 and give your opinion, the attempt to underpower the 6900K, no quad channel during the conference on X99 mobos, the soo called 'Call me before posting your Ryzen review' - Intel

 

[geon2k2]

A derailed Ryzen hype train has brought us these strange things:

• For best performance you suddenly need BIOS updates, games rewrites, etc. etc. etc. Just like with RX 480 which needed tons of work to show its true performance while so far the truth has been that NVIDIA destroys AMD in D3D 11 and earlier titles and AMD wins in some D3D 12 titles by the virtue of certain games being optimized and written for the GCN arch from the get go (read Doom). But even this advantage will soon be partially negated by D3D 12 ready NVIDIA drivers which will substantially improve NVIDIA's performance.
  
• TDP is enormous and even more so when being overlocked. It's nowhere near the advertised 95W. It's still miles better than with previous AMD architectures but the hype train certainly derailed here.
  
• AVX2 performance is halfed thus Ryzen 1800X is practically the same as 7700K in AVX2 workloads (x.265 encoding) even though it has two times more cores and substantially higher TDP.
• I wonder why I've never seen reviews prior to the launch of Ryzen which mentioned micro-stuttering in certain games with the Intel 7700K CPU. Suddenly Ryzen gives better fluidity while it's never been the case before. I call this BS.
• AMD has suddenly decided that most gamers play at 4K. Nope, it's less than 5%.

I was about to dive in but all things considered I will not be upgrading from my Intel Core 2500. Ryzen is an excellent workhorse for certain enterprise workloads but I have a simple desktop under the table. I will wait for Ryzen 2.0 with

• Better overall performance
  
• Reworked AVX 2.0 support (not sure it's possible)
  
• Better clocking out of the box
• Decreased TDP
• Largely improved IMC

or Coffee Lake which will offer six cores on the desktop.

Ryzen launch was certainly rushed but I do understand why AMD did that.

Actually in CPU HASH and AES its crashing intel big time so AVX works just fine:
http://www.hardwarecanucks.com/foru...4-amd-ryzen-7-1800x-performance-review-7.html

Power consumption is pretty decent, in line with enthusiast intel parts while for R7 1700 is just amazing for a 8 core 16 thread part..
Very good idle and amazing full load, even lower than the i7-6700K, which it trashes in everything but the games, and I have the feeling that will also be resolved by R5, with lower core count and higher frequency:
http://hothardware.com/reviews/amd-ryzen-7-1800x-1700x-1700-benchmarks-and-review?page=10

 

[medi01]

German site reported 17% fps increase just from bios update:

Verglichen mit dem ursprünglichen Bios steigert das neue UEFI die Bildrate in unserem Spiele-Parcours zwischen plus 4 und plus 26 Prozent, im Mittel gar um plus 17 Prozent! Angesichts dieses gewaltigen Leistungszuwachses mussten wir Gewissheit haben, dass unsere Werte korrekt sind, und haben mit den Asus-Boards nachgemessen. Damit erreichen wir einen Hauch mehr Geschwindigkeit in Games als mit der aktualisierten MSI-Platine.

Compared to the original bios, the new UEFI increases the image rate in our game course between plus 4 and plus 26 percent, on the average even plus 17 percent! In view of this tremendous increase in performance, we had to be certain that our values are correct, and have measured with the Asus boards. This gives us a touch more speed in games than with the updated MSI board.

https://www.golem.de/news/ryzen-7-1800x-im-test-amd-ist-endlich-zurueck-1703-125996-4.html

...for best performance you suddenly need BIOS updates...

And that for a product that was released WHOPPING day ago.
Pathetic.

Remind me, how often did we have issues with Intel's chips.

 

[CAPSLOCKSTUCK]

@Raevenlord

i am baffled by the "missing chip" business.


1. Was it sent in regular mail or what?
2. Has it/they arrived yet
3. Is there any point now in doing your own review?

 

[medi01]

 

[birdie]

Either someone is an [del]idiot[/del] or just doesn't understand what he's saying.

What does AVX 2.0 have to do with SHA hashing? Who runs SHA hashing workloads on their PCs/Workstations/whatever? You want to mine? Any GPU will be from 5 to 20 times faster while mining than your super duper Ryzen CPU.

Here's one of the examples of AVX 2.0 performance in Ryzen (barely faster than 7700K):

AVX 2.0 is maimed in Ryzen due to the 128bit FMACs vs 256 FMACs in Intel CPUs and there will be no BIOS update for that.

 

[medi01]

AMD_LisaSuCEO of AMD 148 points 48 minutes ago

Ryzen is doing really well in 1440p and 4K gaming when the applications are more graphics bound. And we do exceptionally well in rendering and workstation applications where more cores are really useful. In 1080p, we have tested over 100+ titles in the labs…. And depending on the test conditions, we do better in some games and worse in others. We hear people on wanting to see improved 1080p performance and we fully expect that Ryzen performance in 1080p will only get better as developers get more time with “Zen”. We have over 300+ developers now working with "Zen" and several of the developers for Ashes of Singularity and Total Warhammer are actively optimizing now


But isn't it unfair to compare CPUs if there is GPU bottleneck? Then all CPUs will perform similarly.

AMD_RobertTechnical Marketing
21 points 17 minutes ago

First, I think it's important that readers get a complete picture of a processor. People who have 1440p and 4K displays deserve to read how their potential processor will perform on the monitor they have. Don't you agree? We're also not shying away from the 1080p results. We clearly have some work to do with game developers on some of these titles to invest in the vital optimizations that can so dramatically improve an application's performance on a new microarchitecture. This takes time, but we'll get it done. But what's also clear is that there's a distribution of games that run well, and a distribution of games that run poorly. Call it a "bell curve" if you will. It's unfortunate that the outliers are some notable titles, but many of these game devs (e.g. Oxide, Sega, Bethesda) have already said there's significant improvement that can be gleaned. We have proven the Zen performance and IPC. Many reviewers today proved that, at 1080p in games. There is no architectural reason why the remaining titles should be performing as they are.

AVX 2.0 is maimed in Ryzen due to the 128bit FMACs vs 256 FMACs in Intel CPUs and there will be no BIOS update for that.

True.
That allowed AMD to save quite a bit of space on Zen's silicon.

 

[R0H1T]

Either someone is an [del]idiot[/del] or just doesn't understand what he's saying.

Here's one of the examples of AVX 2.0 performance in Ryzen (barely faster than 7700K):

AVX 2.0 is maimed in Ryzen due to the 128bit FMACs vs 256 FMACs in Intel CPUs and there will be no BIOS update for that.

Clock for clock it's 30% faster than 7700k, that doesn't qualify as barely, & about 20% slower than 6900k, probably down more due to quad channel mem on x99

 

[Caring1]

What does AVX 2.0 have to do with SHA hashing? Who runs SHA hashing workloads on their PCs/Workstations/whatever? You want to mine? Any GPU will be from 5 to 20 times faster while mining than your super duper Ryzen CPU.

SHA isn't just used in Mining, it is an instruction set in the CPU for a reason, many algorithms run using it. Not all can utilise a GPU unless it is especially written for it.
To quote you:

 

[nemesis.ie]

@cadaveca or someone who has one, can you run this during some gaming and see what, if anything appears please:

http://www.thesycon.de/eng/latency_check.shtml

I won't have my kit until next week and may not be able to put it together for a bit after that (recovering from surgery).

Thanks!

 

[birdie]

That's amazing.

Barring highly redundant SHA hashing performance (99.99% of people run SHA hashing 0.0001% of their CPU time), Ryzen (in single threaded tasks) is a tad slower than a 6 years old Core i5 2500:

https://browser.geekbench.com/v4/cpu/compare/38140?baseline=1988008

But I'm positively sure AMD fans will weasel out of this comparison as well.

 

[Camm]

That's amazing.

Barring highly redundant SHA hashing performance (99.99% of people run SHA hashing 0.0001% of their CPU time), Ryzen (in single threaded tasks) is a tad slower than a 6 years old Core i5 2500:

https://browser.geekbench.com/v4/cpu/compare/38140?baseline=1988008

But I'm positively sure AMD fans will weasel out of this comparison as well.

You just compared a benchmark running on two different OS's btw. Great job.

 

[nemesis.ie]

Another question is how much (cumulative) improvement have we seen on the Intel side since the i5 2500?

Overall performance vs price (including running cost potentially) for the work-load(s) you use is all that really matters, the age of the hardware is irrelevant (unless it's dying!).