Ok, one more observation/question. Again (somewhat recent) historically, if I am remembering correctly, AMD comes out with something that about catches up with Intels then current (and usually out for a while at that point) processors in the performance, but within what seemed to be fairly short order, Intel dropped the next bomb (new gen CPU) on them, and Poof went AMD's sales numbers.

So my question would be, does anyone know if Intel has the next version (post Kaby Lake?) ready to drop out of the sky on their heads again, or possibly being able to drive the cost of Kaby Lakes (this is their latest offering, correct?) down enough to cause pain and suffering in AMDville? Again, I'm a bit out of the loop other than what I read around here mostly, as I don't build many latest and greatest systems any more. Thanks for the education!

Yes, this is what Intel has historically done in the market to stymie competition. I believe that next CPU platform after Kaby Lake is Cannon Lake and is scheduled to be released 4Q 2017. So that matches up the scenario you describe of AMD only being able to market their best new CPU for a couple of months before Intel pulls the rug out from under them again. Cannon Lake should push the Kaby Lake prices down to make room for their next product. I have heard rumors though that Intel is having issues getting any releasable yields out of the 10 nm processes they will use on Cannon Lake so the launch date may slip. Intel is also supposed to launch the Coffee Lake processor at 14nm, same as Skylake , following the launch of Cannon Lake. I have a hunch we will see Coffee Lake before Cannon Lake because of its manufacturing issues.

While following up the new leads in this thread, I stumbled on some leaked synthetic benchmarks of the R7-1700X, the chip I am interested in and which is supposed to be the equivalent stepping as the soon to be released commercial product. Tests were run on the mid level B350 chipset motherboard and Turbo wasn't enabled.

first-amd-ryzen-7-1700x-benchmarks-are-here

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Ok, one more observation/question. Again (somewhat recent) historically, if I am remembering correctly, AMD comes out with something that about catches up with Intels then current (and usually out for a while at that point) processors in the performance, but within what seemed to be fairly short order, Intel dropped the next bomb (new gen CPU) on them, and Poof went AMD's sales numbers.

IMO intel would be stupid killing AMD off, to get better they need some sort of competition. And there are those pesky anti-trust laws.

---

Ok, one more observation/question. Again (somewhat recent) historically, if I am remembering correctly, AMD comes out with something that about catches up with Intels then current (and usually out for a while at that point) processors in the performance, but within what seemed to be fairly short order, Intel dropped the next bomb (new gen CPU) on them, and Poof went AMD's sales numbers.

Not really.
Cyrix, then AMD, CPUs were always very much second fiddle to Intel. At their best they would come close(ish) to similar performance for a given clock speed for Integer work, but their Floating Point performance was generally woeful (at best) compared to Intel. Then Intel came up with MMX (often called MultiMedia eXtensions, although apparently it had no official meaning. Just a catchy, copyrightable name), the original SIMD (Single Instruction, Multiple Data) instructions. AMD came up with their own version, and due to all sorts of cross licensing weirdness also implemented Intel's version as well.
But the fact was Intel moved even further ahead performance wise. And AMD still had the same reputation as Cyrix before them, poor performance & run damn hot.

Then Intel moved to the Netburst architecture, along with Rambus DRAM. Around that time AMD came up with the K7 (which become the Athlon) which not only matched Intel's Integer performance, and it's SIMD performance, but out did Intel in FP. It beat their existing CPUs, and it outperformed their newer Netburst CPUs. AMD continued to tweak the design & Intel couldn't match them. AMD actually took a big chunk of the server market from Intel.
That's when Intel went back to the drawing board, dumped Netburst and RDRAM & came up with the Core architecture.

Pretty much since then AMD were relegated to second place. AMD released Bulldozer with all sorts of claims & promises. It didn't deliver. Intel continued to get further ahead with performance, while reducing power consumption and AMD, even with several revisions, continued to fall further & further behind.
Ryzen looks as though it might actually offer some competition.

But I want to see lots of different reviews of different retail chips to find out if that really is the case, or this is just Bulldozer v2. I hope not. I really, really hope not. It looks very promising, but I want to see how it performs in the real world.

---

Grant
Darwin NT

Ok, one more observation/question. Again (somewhat recent) historically, if I am remembering correctly, AMD comes out with something that about catches up with Intels then current (and usually out for a while at that point) processors in the performance, but within what seemed to be fairly short order, Intel dropped the next bomb (new gen CPU) on them, and Poof went AMD's sales numbers.

IMO intel would be stupid killing AMD off, to get better they need some sort of competition. And there are those pesky anti-trust laws.

Didn't stop them last time. Intel got away with coercing all the OEM's from ever even considering using AMD processors or they would get all their incentive money yanked. For a decade, no OEM's would consider building systems or laptops with anything but Intel processors. In the end, Intel got a wrist slap and token fine from the regulatory agencies that was a blip in their bottom line income. So those pesky anti-trust laws don't raise any concerns for Intel. The 800 lb gorilla rules the market. They still have not paid AMD one cent of the $2.5B the courts ordered Intel to pay AMD.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

remember http://www.pcworld.idg.com.au/article/602228/eu-court-has-intel-inside-appealing-1-4-billion-antitrust-fine/ :D

writing from an old system with Xp2800+ , Epox 8rda3+ MB, Radeon 9500 ( 03' ) even run at 2.5Ghz under watercooling in the past
and my Seti crunch box run with a a8n-e and a A64X2 4400+ toledo at 2.5hz with watercoling 24h a day 365day an year .without any error :D

---

remember http://www.pcworld.idg.com.au/article/602228/eu-court-has-intel-inside-appealing-1-4-billion-antitrust-fine/ :D

writing from an old system with Xp2800+ , Epox 8rda3+ MB, Radeon 9500 ( 03' ) even run at 2.5Ghz under watercooling in the past
and my Seti crunch box run with a a8n-e and a A64X2 4400+ toledo at 2.5hz with watercoling 24h a day 365day an year .without any error :D

Yea, but the €1.05B settlement was to be paid to the EU courts, NOT AMD. AMD will not get a penny from Intel from this judgment if and when Intel pays the EU courts. They STILL have not paid the $1.45B judgement the FTC leveled against them. That judgment at least is supposed to be paid directly to AMD. AMD still has not received any of the settlement money from Intel. That judgment was put in effect in 2009 and is still in litigation by Intel. I doubt Intel will ever pay up anybody. They are too big and have too many well paid lawyers. One of the reasons I support the little players in the market. Intel's arrogance greatly aggravates me.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I doubt Intel will ever pay up anybody. They are too big and have too many well paid lawyers. One of the reasons I support the little players in the market. Intel's arrogance greatly aggravates me.

Same as for all the judgements against Apple, AMD, Nvidia, Samsung, Google, IBM etc.
The cases generally drag on for 10-30 years and at the end the parties involved generally agree to call it quits. Even then it's very rare for one party to pay the other even a token amount to wind things up.

---

Grant
Darwin NT

Same as for all the judgements against Apple, AMD, Nvidia, Samsung, Google, IBM etc.
The cases generally drag on for 10-30 years and at the end the parties involved generally agree to call it quits. Even then it's very rare for one party to pay the other even a token amount to wind things up.

True. When it's nothing more than a trade or patent/intellectual property case, this has been the pattern. Mainly because any agency having jurisdiction over the case has no effective enforcement powers. Only when the case is criminal does anything of consequence happen in a timely manner.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Looks like the Ryzen performance benchmarks are leaking in increasing frequency as we approach the February 28th GDC conference event and release on March 02. I see that the reviews of the R5 1600X have slipped out across multiple news outlets. They even got a Forbes mention and that is getting into mainstream media territory. I think that the picture is getting rosier every day now for AMD. Intel should be worried.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Ryzen 7 SKU's and motherboards are available for pre-order today. I am putting together a shopping list. I have some technical questions to ask the Intel crowd for the crunchers running the latest generation Skylake or Broadwell CPU's on motherboards that support M.2 NVMe drives and who are actually running NVMe storage. Can anyone comment on the impact that the NVMe drives have on either/or any degradation in CPU processing power or GPU processing? I doubt there is much effect on GPU processing since we crunchers have pretty much figured out that even X4 PCIe speeds have little performance degradation on our GPU output, but have a question on CPU processing since we do take away X4 PCIe lanes to support NVMe drives. Any comments or observations?

Pretty much decided on Ryzen 7 1700X and ASUS Prime X370 motherboard bundle along with 16 GB Corsair DDR4 3000 CL15 memory but was wondering whether to splurge on a new M.2 NVME drive to replace my existing SATA 600 SSD drives. Hoping someone with the previously mentioned hardware already crunching BOINC projects can comment.

FYI, Ryzen 7 CPU's already top 3 best sellers on Amazon.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Well Keith,

I don't own any but from I thought M2 only supported Win 10?

The only other thing I remember reading is overheating issues with the M2 that slowed them down. So there are now dedicated M2 coolers that you can buy or 3-D print to try and keep the temps down.

You will have to give us a report on how they operate.

Zalster

---

Well Keith,

I don't own any but from I thought M2 only supported Win 10?

The only other thing I remember reading is overheating issues with the M2 that slowed them down. So there are now dedicated M2 coolers that you can buy or 3-D print to try and keep the temps down.

You will have to give us a report on how they operate.

Zalster

From what I understand, all the M.2 form factor drives are PCIe based. The drive manufacturers provide installation software that copies the necessary PCIe storage driver into the Windows installation. Support is available with the required driver or Microsoft Hotfix in Windows 7. There already is native PCIe drive support in Windows 8.1 and 10. Bigger question is whether it really is necessary to run Windows 10 for Ryzen support. I haven't seen explicit information that this is really the case yet. Skylake is supposed to be only supported in Windows 10 yet there are Youtube videos showing Skylake processors running on Windows 7.

One of the reasons I have decided I like the ASUS x370 Prime motherboard is that is doesn't have the M.2 location directly underneath a hot graphics card. That is what I have seen in the majority of the X170 and X270 Intel motherboards. Not a good location for a hot running M.2 drive and there can be throughput issues with the drive if it decides it needs to throttle down because it is running hot.

The biggest roadblock first is the motherboard BIOS needs to support the NVMe drive and it appears all the new X370 and B350 chipset boards do with their UEFI BIOS'

My decision is whether to just move my existing components over to the new motherboard/CPU/memory hardware along with the motherboard drivers of course. Or whether to try and clone my existing SSD AHCI boot drive over to a new M.2 NVMe drive. I like the idea of eliminating one more set of cables from the system.

If I go that route I will post my tribulations to this thread.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

All the Ryzen information released to day looks promising, but I still want to see independent benchmarks of retail products. Particularly single threaded performance.

---

Grant
Darwin NT

All the Ryzen information released to day looks promising, but I still want to see independent benchmarks of retail products. Particularly single threaded performance.

I'm taking a leap of faith that what has been shown in the leaks is verified by 3rd party reviewers. The Cinebench R15 single-threaded performance was matched exactly with the Ryzen 7 1800X CPU versus the i7 6900K CPU. We shall see in 8 days I guess.

Just put my pre-order in for the motherboard/CPU/memory. Memory shows up Saturday and will wait on the rest till next week I guess. Didn't do anything with the M.2 drive. Will cross that bridge when necessary. These parts are going to go into my slow Windows 10 system. Don't know what to do until March 2 regarding any Corsair H100iV2 AM4 adapter necessary to install everything. Corsair is holding their cards close until then other than saying there will be upgraded parts available on that date.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Hmmm, certainly is starting to look compelling. Might be time to dust off the compilers and mutithread the CPU apps.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Hmmm, certainly is starting to look compelling. Might be time to dust off the compilers and mutithread the CPU apps.

Of all the leaked CPU performance tests, Ryzen beat out the equivalent Intel CPU in 6 of 8 tests. What I found interesting is the R7 was considerably slower in the Prime Number factoring test. I wonder how can that be when it was considerably faster in the Integer and Floating Point tests. Does prime factoring NOT involve Floating Point? I thought it did based on my observations of the AIDA64 and OCCT Prime stress tests. Or is it just that the Prime test is optimized to Intel compiler or something?

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Any unbiased reviews out there yet?

Nope?

Then I'll just have to wait for a 2500K replacement that'll do more work for less power a while longer yet I guess. :-(

Cheers.

Hmmm, certainly is starting to look compelling. Might be time to dust off the compilers and mutithread the CPU apps.

Of all the leaked CPU performance tests, Ryzen beat out the equivalent Intel CPU in 6 of 8 tests. What I found interesting is the R7 was considerably slower in the Prime Number factoring test. I wonder how can that be when it was considerably faster in the Integer and FFT tests. Does prime factoring NOT involve FFT? I thought it did based on my observations of the AIDA64 and OCCT Prime stress tests. Or is it just that the Prime test is optimized to Intel compiler or something?

Unsure on that. A cursory look at algorithms for prime factoring says they vary widely. While i don't immediately spot explicit FFTs involved, they could be being used as a black box in some of the tests.

For problems of any reasonably large size like that, we'd probably have to find out what sortof memory configuration was used.In the few where I have seen it specified timings were loose and speed lackluster. If libraries with run time optimisation (such as fftw) are used, then it'll pick fastest code paths (i.e. 'wisdom'), however given that we're supposedly looking at a clean slate design, it's unlikely that optimal codelets exist yet in bench software.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Any unbiased reviews out there yet?

Nope?

Then I'll just have to wait for a 2500K replacement that'll do more work for less power a while longer yet I guess. :-(

Cheers.

I'm not sure what you mean by unbiased. The leaked performance reviews I've been reading are not by AMD but by third party testers, who I admit are not mainstream. Whether they are unbiased, I don't know. I'm not aware of any AMD affiliations with those entities. I'm guessing you are referring to reviews by mainstream reviewers like PCPerspective, AnandTech,Tom's Hardware, JaysTwoCents and Paul's Hardware and such?

I just get the feeling that there has been a bit of 'wink-wink' when posters have interrogated those reviewers and get the NDA party line response. And a lot of hold off on buying anything till March 2.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Unsure on that. A cursory look at algorithms for prime factoring says they vary widely. While i don't immediately spot explicit FFTs involved, they could be being used as a black box in some of the tests.

For problems of any reasonably large size like that, we'd probably have to find out what sortof memory configuration was used.In the few where I have seen it specified timings were loose and speed lackluster. If libraries with run time optimisation (such as fftw) are used, then it'll pick fastest code paths (i.e. 'wisdom'), however given that we're supposedly looking at a clean slate design, it's unlikely that optimal codelets exist yet in bench software.

I never have seen a 'wisdom' in either Prime95 or OCCT. But it might be named something different that does the same optimization. I have used those primarily to stress test overclocking settings on my AMD systems to determine the highest stable overclock with a reasonable temperature and power outcome.

Will be interesting to see the Ryzen 7 Anandtech tests next Thursday. I've always like them because they do a series of Math tests also and not just the exclusive business app or gaming tests that most reviewers do.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)