AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

doubtful theyre gonna enforce that on a large scale. probably just covering their backs legally

 

AMD and Intel's Rules on Warranty Coverage are about the same, and both have been doing this for many years, it's nothing new. Box coolers only, no Overclocking or running outside stock specifications.

I only mention it now so the newbies are aware. Otherwise we get accused of recommending coolers and solutions that void the CPU warranty - it's SOP every new CPU generation release.

 

You forgot to add in that Intel dont offer or better say force you to use stock coolers for their K processors anymore. Use what you want/have, because you won't void the warranty due the cooler or fans you prefer to use

 

(Stock speeds) Ryzen 5 2600 vs Ryzen 5 1600 Test in 8 Games

Spoiler

Published on Apr 24, 2018
AMD Ryzen 5 2600 vs Ryzen 5 1600 in 8 Games (GTX 1080)

Games:
Project Cars
Fallout 4 - 01:16
The Witcher 3 - 02:42
Grand Theft Auto V - 04:00
Far Cry 5 - 04:55
Assassin's Creed Origins - 05:57
Battlefield 1 - 07:32
Arma 3 Apex - 08:54

System:
Windows 10
AMD Ryzen 5 2600 3.4Ghz
AMD Ryzen 5 1600 3.2Ghz
Gigabyte X470 AORUS ULTRA GAMING
GTX 1080 8Gb
16Gb RAM 3200Mhz

(OC) I7 7700K vs (OC) R5 2600 | Comparison |
For Gamers
Published on Apr 23, 2018

 

AMD launch low-power 35w Ryzen 3 2200GE and Ryzen 5 2400GE APUs
Damien Mason 6 hours ago APU, Featured Tech News
https://www.kitguru.net/tech-news/f...power-ryzen-3-2200ge-and-ryzen-5-2400ge-apus/

AMD has bolstered its line of APUs, equipping the brand new Ryzen 3 2200GE and Ryzen 5 2400GE chips with Vega graphics technology. While the new processors don’t offer much more in terms of raw performance, the focus this time around is power efficiency as both offer nearly half the thermal design power (TDP) of their predecessors.

Following on from the Ryzen 3 2200G and Ryzen 5 2400G, AMD’s confusingly titled Ryzen 3 2200GE and Ryzen 5 2400GE processors reduce the TDP from 65W down to just 35W. This comes at the sacrifice of clockspeed, which is slightly lower than their predecessors while retaining the same number of cores and threads.

The reason for this sacrifice is power efficiency. Although AMD is currently marketing both APUs as desktop solutions, inevitably the technology will be implemented into laptops and other portable devices, where it’s best suited to help reduce the drain on battery life.

Neither APU is built with highly demanding tasks in mind, however, it’s expected that the graphics-infused processors will be included in small form factor PCs, enabling users to complete their day-to-day work without the need for a dedicated graphics card. This applies to some games, although settings will vary with regards to performance.

Release date and pricing have yet to be confirmed, but AMD’s GE chips are expected just above £100 when they hit shelves. Elsewhere, AMD has its 2nd generation Ryzen CPUs out on the market including the highly impressive Ryzen 5 2600X and Ryzen 7 2700X."

AMD launches Ryzen 5 2400GE and Ryzen 3 2200GE, low-power APUs
Published: 23rd Apr 2018, 09:06 GMT
https://videocardz.com/76013/amd-launches-ryzen-5-2400ge-and-ryzen-3-2200ge-low-power-apus

" AMD Ryzen 2000
Last week AMD introduced Ryzen 2000 (Zen+) processors, which were followed by low-power APUs.

Low-power Ryzen APUs

The GE series are basically 35W TDP version of the 65W APUs. Both Ryzen 5 2400GE and Ryzen 3 2200GE are clocked lower than their 65W variants (~300 MHz difference in base frequency). The clock speeds of the integrated graphics are the same, but we are not sure if what are the other limitations here since 35W have to be shared between CPU and GPU.

Despite the fact that both APUs are listed as ‘2nd Gen AMD Ryzen’, the specifications seem to suggest otherwise. According to the specs at AMD website, low-power APUS are using 14nm FinFET technology, while Zen+ CPUs are 12nm.


Pricing has not yet been confirmed."

 

Intel has not included a box cooler on unlocked -K CPUs since Skylake, so the same rules don't apply.

 

Nah, nobody's gonna doubt this one...

4.8ghz 2700x?
Captin Camerica
Started streaming 3 minutes ago

Spoiler

GPU: GeForce GTX 1070
CPU: AMD Ryzen 7 2700X Eight-Core Processor
Memory: 16 GB RAM (15.46 GB RAM usable)
Current resolution: 1920 x 1080, 60Hz
Operating system: Microsoft Windows 10 Home

Hmm, looks like FutureMark made another typo entering data into their CPU database again - man, I've reported IDK how many times over the years wrong data, and they keep on doing it, Ryzen "5" 2700x isn't correct.

Here's his run:
https://www.3dmark.com/sd/5082520

He might not be best performance stable on 4.8ghz, as someone (maybe him?) got mostly better results at a lower clock rate on Ryzen 2700x:

https://www.3dmark.com/compare/sd/5080895/sd/5082520

 

How To Run (Update BIOS) Ryzen 2700X 2600X 2700 2600 On X370 ASUS
Lawrence Timme
Published on Apr 24, 2018
Use a Ryzen 1000 series to update the bios on your X370 to allow you to upgrade to the Ryzen 2000 series on an Asus motherboard.

 

His Sky Diver physics score is crippled. Picked only some scores. Not the highest https://www.3dmark.com/compare/sd/5082520/sd/4796320/sd/4714500#

 

Ryzen 5 2600X vs i5 8600K | Tested in 5 Games @ 1080p
theTechnician

Published on Apr 24, 2018
Intel Core i5 8600K vs AMD Ryzen 5 2600X Tested in Battlefield 1, Far Cry 5, GTA V, Rise of the Tomb Raider & The Witcher 3 at 1080p.

Spoiler

If you guys are new here then make sure to hit that SUBSCRIBE button for more stuff like this & help me reach 5K SUBS ➤ https://goo.gl/Dfe2b1
▷▷▷▷▷▷▷ TEST BENCH SPECS ◁◁◁◁◁◁◁
♦ CPU 1 - Intel Core i5 8600K @ 4.3GHz
♦ CPU 2 - AMD Ryzen 5 2600X @ 4.2GHz
♦ CPU Cooler - Corsair H100i
♦ GPU - Zotac GTX 1070 AMP! Edition
♦ RAM - G.Skill Trident Z 16GB DDR4 3000MHz (2x8)
♦ Mobo 1 - Gigabyte Aorus Z370 - Gaming 5
♦ Mobo 2 - Gigabyte Aorus AX370 - Gaming 5
♦ PSU - Corsair CX 750M
▷▷▷▷▷▷▷▷▷ VIDEO INFO ◁◁◁◁◁◁◁◁◁
► FPS Monitoring : MSI Afterburner
► Gameplay Recorder : Nvidia Shadowplay
★ NOTE : There's always a 2-5 FPS hit when using an On-Screen gameplay recorder, so keep that in mind.
▷▷▷▷▷▷▷ CONNECT WITH ME ◁◁◁◁◁◁◁
➊ Facebook ➢ https://www.facebook.com/theTechnicianYT
➋ Twitter ➢ https://twitter.com/theTechnicianYT

 

Can you overclock with the AMD Prism air cooler?
JayzTwoCents
Published on Apr 24, 2018
When AMD released the Wraith cooler it was widely accepted as a decent stock cooler... what about AMD's new Prism cooler found bundled with the 2700x??

 

Wow, I hope this is true:

" Update Tuesday 24th:
We got word back from AMD, and much of the staff is as surprised to find out what is written there. AMD mentioned that the FAQ and info distributed is old (5 years) and is being updated soon to clarify things. "

Your Ryzen processor warranty states you may only use the stock cooler (updated)
by Hilbert Hagedoorn on: 04/23/2018 07:27 PM
http://www.guru3d.com/news-story/yo...states-you-may-only-use-the-stock-cooler.html

 

https://www.anandtech.com/show/12678/a-timely-discovery-examining-amd-2nd-gen-ryzen-results/5

A whole lot of words to say we were wrong, and 8700K in indeed faster in gaming. That took way too long IMO.

 

'bout damn time

 

Exactly... "Moving forward, the immediate goal here at AnandTech is to ensure that our readers have the most up-to-date and correct results, particularly for our Ryzen 2000-series review. As a result, we are taking a few steps both immediately and in the future to correct our data, update our Ryzen 2000-series review, and to prevent this issue going forward."

 

Ha, made me check my own system to make sure HPET not forced to On in my operating system:

"bcdedit /enum" in the command line & then checking to see that "useplatformclock Yes" is not listed amoung the output of lines - courtesty of the following thread:
https://www.reddit.com/r/Planetside...d_off_hpet_in_windows_10_now_i_have_an_extra/

 

And here's the quote for @hmscott, who was patiently waiting for Anand to update their reviews, but I cannot find that thread...

From the article - "As a result we are retracting our existing results for all of the processors we used in the Ryzen 2000-series review. This goes for both the review and for Bench."

 

Massive Overkill: ASUS Crosshair VII Hero X470 Motherboard Review
Gamers Nexus
Published on Apr 25, 2018
We review the ASUS Crosshair VII Hero X470 motherboard for its VRM and overclocking quality with Ryzen 2 CPUs.
The ASUS Crosshair VII Hero X470 motherboard is one of the most "overkill" that ASUS has made ever, leveraging 60A power stages for Ryzen 2 CPUs, e.g. the Ryzen 7 2700X. We cover overclocking features on the ASUS C7H X470 motherboard, VRM design and phase count, PCB design and layout, and more. This review of the ASUS Crosshair VII Hero X470 should help you decide if it's "worth it" for Ryzen PCs, and is analyzed by a professional overclocker. We've previously looked at the Gigabyte X470 Gaming 7 vs. Crosshair VII Hero, which you can find linked below (on their youtube channel).

 

Actually Anandtech were right the first time. And, the results should stand as published.

If it weren't for Intel's vulnerability mitigations, no change would have been noticed from previous reviews using Anandtech's testing tools.

The fact that Intel now looks ridiculously encumbered by their mitigations is a wonderful side effect of Anandtech's testing methodology.

I think Anandtech should keep their published results up, and point to the problem they found, which is that Intel is in trouble.

http://forum.notebookreview.com/thr...o-build-a-desktop.815996/page-3#post-10717916

I have a few posts there, but that sums it up.

Intel shouldn't get a pass on this, and Anandtech shouldn't change their test suite.

And, just the same I don't think multiple test runs by reviewers - with mitigations, without mitigations, with HPT timers (which ever flavor is required) on and off - should be required.

We need to be made aware of problems like this, that is what reviewers are supposed to find and illuminate.

Saying that Intel should get "tweaks to improve performance", is favoring Intel over AMD, and in reality - for all of us, it should be the other way around.

Intel needs to come out with a new architecture CPU as soon as possible, it's not funny any longer, their CPU's are broken, and the "spare / mitigations" patched on isn't going to last long.

Wow, look at the hit the mitigations take when additional system call load is added when running high CPU usage applications, it's amazingly disastrous. That's why datacenters have been turning off the mitigations and crossing their fingers, they'd have to double their server farm sizes just to stay even against that kind of performance loss.

(you can click on the graph to go to the article)
Anandtech Article Comments

 

AMD has done this for years. Just send the cooler back with the cpu, and you will have no problems with warranty. I believe the cooler has a matched serial number with the cpu.

 

It counts on if HPET is a requirement for the security mitigations. If not an integral component then disabling it should not be a problem. In either case it should be pointed out on each review where it is disabled and the subsequent consequences.

 

It was not this that was my point. The point is what was written in their website. They can come up now and say it was old and not valid... I don't trust a ****y of this. But they jump after Intel as you can see... Because they hadn't other choices. It was too stupid. And Finally they undestood own stupidity by themself.

More 2nd gen Ryzen reviews...

• AMD Ryzen 5 2600X i2Hard (ru)
  
• AMD Ryzen 5 2600 TechSpot
  
• AMD Ryzen 7 2700 Cowcotland (fr)
  
• AMD Ryzen 7 2700X and Ryzen 5 2600X Hardware.info

• AMD Ryzen 7 2700X and Ryzen 5 2600X DarkTech Reviews (ar)

 

It's a bit not that...

HPET wasn't required for the mitigations, it was turned on by Anandtech in their test suite.

Having HPET on illuminated the problem Intel has with an increase in system calls during high CPU usage situations - high CPU utilization in games at 1080p illuminated the problem.

These results also proved that AMD's claim that the mitigations don't affect their performance are true.

Both very useful facts to have come out, simply from running gaming benchmarks.

The realization this should bring about should have far ranging affects, and there will be a lot of System Admins checking their precision timers settings on their Intel servers - if they haven't started already.

The HPET probably was enabled long ago for the test suite as over time the common thought about enabling / disabling HPET has changed - back and forth, but it shouldn't really cause an issue on modern hardware - it's all way fast and shouldn't feel a performance hit from additional system calls.

There are considerations all over the place for enabling / disabling HPET, but they are specific to configurations and really shouldn't come up - and if they do it would be the kind of set up where someone knowledgeable would have already set it right.

The impact on Intel CPU's is all from system call increases under CPU load, and that can come from any number of sources, not just the HPET being enabled - although that can and will happen in professional and datacenter environments often enough so as to be a serious issue.

If Intel's mitigations didn't required bookending system calls Anandtech's situation would have never happened, and we would all be happily reviewing results unaffected by the HPET being enabled, as has been the case for many years prior.

As comments to the article point out, what are the other system settings you aren't telling us about... and truly that is an almost endless list.

To list out all the system, BIOS, application, and set up differences for the environment would be extensive and illuminating, but exhaustive, which is why noone lists them.

For the current Intel CPU issues, and software that is loaded and running at the time of the benchmark runs that adds system calls over that of the average users system would need to be listed as well, not just the impact of the system calls for HPT flavors.

I think people should run their systems as they have been up until Intel's problems started, and not change them now for Intel's sake.

If Anandtech changes their settings to help Intel, it's going to skew all the previous results too, that's a big loss. Why should Anandtech do that?

 

lol, relax @Papusan bro! I was just informing people that are not aware of how AMD does their warranties. I have never installed a stock AMD cooler, on any AMD cpu I have owned, and never had an issue with warranty replacement. I guarantee, if I cook this 2700X, without its stock cooler on it, I will still receive a shiny new CPU from AMD

 

I read it as they were forcing HPET usage for all timers in Windows. By default Windows uses it at times at others it will not. What they seem to be doing is to allow the OS to chose the timer rather than force HPET. This is the default behavior as most users will have it on their systems. To me that just means they are correcting an errata that they have caused. Just so long as they fix the errata on both systems, even if the affect is minimal.

 

Interestingly AMD Ryzen 1.0 recommended HPET off as it produced a 5% to 8% performance hit.

For Ryzen 2.0 AMD has reversed that recommendation as AMD has fixed the performance issues and recommends that HPET be enabled always in the BIOS - and AMD suggests to not allow HPET to be disabled - so the BIOS may not even have an entry to change it from the default of Enabled.

If the BIOS enables HPET and the OS disables HPET, then HPET is off.

Anandtech details as such here:

A Timely Discovery: Examining Our AMD 2nd Gen Ryzen Results

AMD and Intel Have Different HPET Guidance
https://www.anandtech.com/show/12678/a-timely-discovery-examining-amd-2nd-gen-ryzen-results/3

Historically Anandtech was ok enabling HPET as it didn't cause performance issues. Now, Intel has a problem with increased system calls due to the mitigations:

" Intel
The response we had from Intel was a little cryptic:

[The engineers recommend that] as far as benchmarking is concerned, it should not matter whether or not HPET is enabled or not. There may be some applications that may not function as advertised if HPET is disabled, so to be safe, keep it enabled, across all platforms. Whatever you decide, be consistent across platforms.

A cold reading of this reply would seem to suggest that Intel is recommended HPET to be forced and enabled, however my gut told me that Intel might have confused ‘on’ in the BIOS with ‘forced’ through the OS, and I have asked them to confirm.

Looking back at our coverage of Intel platforms overall, HPET has not been mentioned to any sizeable degree. I had two emails back in 2013 from a single motherboard manufacturer stating that disabling HPET in the BIOS can minimise DPC latency on their motherboard, however no comment was made about general performance. I cannot find anything explicitly from Intel though.
...
Based on my extreme overclocking roots back in the day, my automated benchmark scripts for the past year or so have forced HPET through the OS. Given that AMD’s guidance is now that it doesn’t matter for performance, and Intel hasn’t even mentioned the issue relating to a CPU review, having HPET enabled was the immediate way to ensure that every benchmark result was consistent, and would not be interfered with by clock drift on special motherboard manufacturer in-OS tweaks. This was a fundamental part of my overclocking roots – if I want to test a CPU, I want to make certainly sure that the motherboard is not causing any issues. It really gets up my nose when after a series of CPU testing, it turns out that the motherboard had an issue – keeping HPET on was designed to stop any timing issues should they arise."

You probably want to read the rest on the site:

Forcing HPET On, Plus Spectre and Meltdown Patches
https://www.anandtech.com/show/12678/a-timely-discovery-examining-amd-2nd-gen-ryzen-results/4

 

HPET is not the default timer used by Windows for Intel systems, and is a fall back timer from what I've read. It comes with performance penalties which is why with the introduction of the Core-i CPUs Intel has been using Invariant TSC with HPET being a fall back if Invariant TSC not being available or what I've read. Ananadtech has been doing it wrong all along and forcing HPET ON which is not how 99% of users will be using their system as Windows defaults HPET OFF. Saying Anandtech should be messing with the OS and forcing an outdated timer to make AMD CPUs look better is nonsense.

According to Microsoft HPET was used during circa Windows Vista and Windows Server 2008... HPET according to them is a fall back if the event TSC is not available which is why it's still available in the OS and BIOS. If you read the Ananadtech the reviewer references how he's always used HPET since his early overclocking days, and based what he was taught. This sounds like a long explanation from Anandtech to say we F'ed up and don't want to own that mistake fully.

I'd be more pissed off if I had been a consumer looking for a new CPU and saw their article and the grossly wrong and misleading gaming benchmarks, only to now have those fixed almost a full week later in favor of the competition. Actually it brings up another question if this had been their intention all along. How do we know they weren't paid a little sum of cash to present what they did, keep it up for a week or so, and then change their review. It certainly didn't hurt sales having it in favor of AMD.

 

I don't think 99% of the users know what is used on their systems

As long as Anandtech implemented the same HPT choice on all the test systems, that should be enough for an even playing field.

Again, without the Intel vulnerability mitigation situation, we would have never seen HPET causing this impact on Intel CPU performance, the same as we aren't seeing any impact on an AMD CPU now.

It's lucky this has happened, so as to illuminate the performance issues inherent with the Intel mitigations, and the corresponding lack of performance impact on the AMD CPU under the same test conditions.

 

Yes. The *default* value should be the one used, as you mentioned 99% of users wouldn't even know how to change it.

In regards to the same setting value, it doesn't necessarily have to be the same, but rather the default. It could be that Windows optimizes settings differently based on CPU, so it's theoretically possible the default value of one is to the detriment of the other.

Making sure default values are used is the best policy in order to closely reflect what end users will see. And if defaults are changed, they should explicitly say why a config value was changed, and provide numbers of both before and after.

 

I found my systems (Ivy and Skylake) much less snappy, micropauses etc. after I was forced to turn HEPT on to bench GPUPI

Wasn't the Ryzen 1 sleep bug/benching cheat the main reason it's even known about?

 

If the OS changes the timer used due to the CPU, wouldn't that be changing the test environment to improve performance? Is that allowed? Is that fair?

If the default favors one brand over another, is that fair too? If the application monitoring benefits from higher precision timing being available, shouldn't it be made available?

If the playing field is the same, then the differences are of significant interest, at least that is how I see it.

I've run with HPET on as default for many systems, and haven't had issues, just like Anandtech's testing over the last couple of years.

If I didn't specifically test the HPET performance impact again after the mitigations were applied, and I ran my standard defaults on a new Intel CPU system with HPET enabled, is that fair to me to not have reviewers test with HPET enabled so I am alerted to the impact in games?

It can certainly be argued both ways, but your way we all stay in the dark, Anandtech's way we find out new things we didn't know before. I like Anantech's way better.

Moving forward if Anandtech bows to the Intel masses to show Intel in it's best light, that's to be expected, but I hope somewhere in the way forward broader testing includes settings that don't favor Intel over AMD, as that's how it feels to me right now.

Intel is getting a pass - hiding from the bad news, when we should instead be paying attention to what has been discovered - and how it wasn't there 6 months ago - but will be with us, the owners of the broken Intel CPU's, forever moving forward.

 

I respectfully agree that we disagree. In general, I would hope all CPU vendors provide an OS author/company all the settings that make their hardware run in the best possible fashion. So I don't necessarily expect configurations to use the same defaults. And if the defaults are changed, then full disclosure is warranted.

 

It was a Windows RTC bug that caused benchmarks returning from sleep to report incorrect results, and the solution was to enable HPET - which needed to be on for Ryzen Master anyway.

Besides, what good benchmarker would wake up a sleeping benchmark and let it continue to completion?, and use that score? No one I know. I would simply kill it and start over, after changing my sleep settings so that doesn't happen again.

hwbot also recommended enabling HPET:

Ryzen Platform Affected by RTC Bias; W8/8.1/10 Not Allowed on Select Benchmarks
http://hwbot.org/newsflash/4335_ryz...bias_w88.110_not_allowed_on_select_benchmarks

"In a statement issued by the Head of Moderation Christian Ney, we have confirmed that the AM4 platform is affected by the Windows 8/8.1/10 RTC bias. The bias occurs when adjusting the reference clock at run-time and will affect the Windows timer, causing benchmarks to perceive time slower (or faster) than it really is. This results in benchmark scores affected in such a way that the benchmark scores reported do not reflect real performance.

The RTC bias is referenced in the ROG Crosshair VI Hero Extreme Overclocking guide available on Overclocking.guide: "Timer is skewed when changing REFCLK in Windows 8+. Additionally the default systimer has issues with OS ratio changes unless HPET is enabled. To summarize, always enable HPET on this platform.""

Hwbot is recommending / requiring benchmarkers to *always* enable HPET.

Now, because Intel gets bad performance results with HPET enabled we are supposed to ignore AMD, Intel, Hwbot advice and for Intel's sake disable HPET?

Some comments in the Anandtech article point out that how do we know that with HPET off that Intel isn't benefiting from some error in timing measurements as well as not seeing the performance hit of the Intel vulnerability mitigations?

Anyway, I am expecting that more research goes into HPT's on Intel with mitigations installed and enabled, to see how much of a performance hit each HPT option experiences, and how the accuracy of timing measurements is affected by the settings.

 

AnandTech is saying in bios to leave it on. There is a command

Code:

bcdedit /set useplatformclock true

that forces HPET to be used all the time. By default the OS should have

Code:

bcdedit /deletevalue useplatformclock

. This command allows Windows and/or the program to chose the timer used.

At least this is my take on all of this...............

 

In reality, there is likely no choice to disable HPET in AMD Ryzen BIOS, and if there is the default there is Enabled. Per guidance from AMD / IBM it is recommended Enabled in BIOS.

Anandtech has in the face over overwhelming whining given in to the mob and disabled HPET and are re-running the tests to get favorable results for Intel CPU's with vulnerabilities installed and enabled.

I hope Anandtech doesn't "erase" the original scores, but leaves them in for comparison - it's good to know how HPET and other high system call operations affect high CPU utilization. I will be disappointed in Anandtech if they cave in on this too and erase the evidence of their discovery by removing the original results - but I will understand why they felt compelled to do so.

Here is Anadtech's parting shot:

"Finally, while we're on the subject of timers, we'd like to throw out an open-ended question to everyone: given what we've found, should the use/requirement of HPET in software be made clearer? Or is there a risk that information being more confusing than helpful? One of the issues we grappled with in writing this article is that while HPET can have a performance impact, it's also not necessarily wrong to use it given its unique accuracy. So we're interested in hearing from all of you on how you think the use of HPET should be documented, so that users aren't caught off-guard by the potential performance impact.."

 

LOL, the coders know what impact their code will have. It is up to then to decide on using high performance code or not. Just leave it at default and be done with it. I find it funny though the all the software over time that was optimized for Intel and not AMD never got this attention but now that the shoe is reversed.

 

Yes, some developers and most System Architects do know, and as I said it depends on the OS, application and tools used, CPU generation used, hardware peripheral considerations, and varies from project to project.

Testing and benchmarking needs to be done to determine the best BIOS and OS configuration for many settings for each project, not just HPT timers, even if only to verify expectations are still in effect.

In Anandtech's case the testing came up positive for Intel CPU sensitivity to high CPU load being affected by system call increases due to high precision timers.

Both AMD and Intel have had trouble with HPT enabled over time varying with the CPU release.

You can't just poo-poo Intel's broken CPU failure situations, such as this one. Intel mitigations are really affecting performance and ruining forecasts for server expansion plans and running current capacity to the limits.

For many it's a matter of gambling on running without protection, or spend a bunch of money on broken architecture CPU's, which will move from not cost effective to worthless after the fixed CPU's are released.

Or, begin the move to AMD so you can plan effectively for capacity and growth without the uncertainty of future exploits requiring additional overhead, further reducing capacity. Assuring capacity value moving forward.

It's really not a funny situation. Get serious, wake up, it's real.

 

Na, not really. Just go by your old guidelines as to what system you needed before and purchase one capable of 35% better performance and you are golden for mitigation performance loss. Good thing too with Ryzen as competition that Intel has now substantially lowered the cost of those systems as well. But true too now we have Epyc as well.

 

Really, customers should spend another $3.5m on a $10m Intel server deployment? Or worse, spend an unplanned $3.5m + contingencies on an unplanned capacity expansion?

I don't think a single customer would enter that path given that information up front.

Especially knowing that Intel is planning to release "fixed" CPU's the following year, and the broken CPU legacy investments, old and new, will be "worthless" after the new CPU's are released.

The next statements would like be something like this:

"We've heard that AMD now has a substantially more cost effective Datacenter solution that isn't performance compromised, would that be a better investment for the additional $3.5m capacity expansion fund you are requesting?"

"We don't have much more expansion space left in the DC's to allow for 35% more space allocation, that wasn't considered necessary until Q3|20.

Additional space on short notice will cost a 75% premium, not to mention all of the additional support hardware - cooling, network, monitoring and control, and none of those departments have budget allocated"

"We could fund a whole new AMD wing on that extra 35% + DC contingencies money. Why give it to Intel, the CPU with the problem? What if we give the money to AMD and let's start a relationship now, before it is too late."

"If Intel's new offering succeeds we can talk about including a limited "Test" Replacement project for Intel in Q2|21, or later if it proves problematic."

"Besides, it's likely better to assume a 75% hit to Intel capacity, and plan for that, 35% doesn't take into account additional exploits and fragmentation per service requirement given the variable performance hit across the DC. We don't want to get caught out on short notice again."

"That means AMD gets an even larger DC Wing."

 

Ryzen 5 2600X vs. Core i7-8700K IPC Comparison, AMD's Hot On Intel's Heels!
Hardware Unboxed
Published on Apr 26, 2018

Gigabyte Aorus X470 Ultra Gaming Unboxing & First Looks
Tech of Tomorrow
Published on Apr 26, 2018
The Best Priced AM4 Motherboard? Gigabyte Aorus X470 Ultra
ASRock X470 Taichi Ultimate Review - What's Different Vs. The X370 Taichi?
Tech YES City
Published on Apr 26, 2018
ASRock have gone away with the fatal1ty line up of motherboards and instead replaced it with the Taichi Ultimate Series (at least for the X470 line), so what are the differences between this and the X370 Taichi? And is it worth the premium? Let's find out!

 

Ryzen 5 2600x vs Core i7 8700k Test in 8 Games
Testing Games
Published on Apr 26, 2018

Spoiler: Details...

AMD Ryzen 5 2600x vs Intel Core i7-8700k in 8 Games (GTX 1080)

Games:
Far Cry 5
Project Cars 2 - 01:00
Assassin's Creed Origins - 02:18
Fallout 4 - 03:57
Grand Theft Auto V - 05:25
The Witcher 3 - 06:21
Arma 3 Apex - 07:51
Battlefield 1 - 09:03

System:
Windows 10
Intel i7 8700k 3.7Ghz
Asus ROG STRIX Z370-H
AMD Ryzen 5 2600x 4.3Ghz
Gigabyte X470 AORUS ULTRA GAMING
GTX 1080 8Gb
16Gb RAM 3200Mhz

(OC) R7 2700X | X470 vs X370 vs B350 | Comparison |
For Gamers
Published on Apr 26, 2018
Ryzen 7 2700 vs Ryzen 5 2600 | Comparison |
For Gamers
Published on Apr 25, 2018
B350 vs X470 | Ryzen 7 2700 | Comparison |
For Gamers
Published on Apr 25, 2018
(STOCK) R7 2700 vs I7 8700 | Comparison |
For Gamers
Published on Apr 25, 2018

 

Agreed!!

It is my understanding there are two things in play here:

1) HPET being enabled in BIOS
2) The Windows configuration for handling of the platform clock (to use HPET, which has to be enabled in BIOS, or use something else)

In Windows 10 (as that is something I could verify), regardless of BIOS setting, the default is to use a "Time Stamp Counter" (tsc) for tracking time, correcting drift, etc. However, from what I read, Anand's scripts change this clock value within Windows to force HPET usage (away from the defaults). This is my sticking point in their methodology and should probably invalidate their results.

BTW, apparently the whole HPET vs. TSC debate has been going on for a while (below is a 6 yr old thread). In there it was recommended to test and configure different settings for timers within your own system as different configurations seem to generate different results depending on the hardware - https://www.neowin.net/forum/topic/...n-bios-and-os-for-better-performance-and-fps/

 

2nd Gen AMD Ryzen™ Processors: Built for Enthusiasts, by Enthusiasts
Published on Apr 26, 2018
Join James as he discusses why, as an Enthusiast, he is excited about the new 2nd Gen AMD Ryzen™ CPUs

 

Wait, so you have no problem when it caused the Ryzen 1.0 deficit in speed, but now that it disfavors your team, you complain and say it was intentional. Let me tell you a story. It is about CPU benchmarks using HPET because of bugs by both competitors at different times over the years. So many benchmarks required HPET to be enabled. We even still have this on some of the benchmarks in HWBot. So, his experience of having it on is relevant to explain intention, cutting across you're statement, and cutting across the idea he meant to harm Ryzen 1 in the benches. It was evenly applied in all his work until now.

Now, we can question the wisdom of his decision as the benches moved away from this requirement of HPET and whether it is reflective of real use (many synthetic benches do not reflect real world performances in one way or another, but can give an idea of performance). That is fair. But much of the talk is just people being upset on all sides.

It is fair to question why change it now when it harms Intel so much, but the correction in settings moving forward should be done sometime to give better into to his readers and maintain the trust of his readers. Overall, this is much ado about nothing.

Sent from my SM-G900P using Tapatalk

 

good to have u back ajc

Sent from my Xiaomi Mi Max 2 (Oxygen) using Tapatalk

 

Excuse me buddy when did I say anything about Ryzen 1 or it's speed deficits or previous issues? Thank you for assuming such things though.

This isn't about favoring or disfavoring a team, its about altering the OS and forcing a non default state which introduced a huge performance issue with Intel system as they don't use HPET as the default config. Anandtech even went as far to use 2666mhz with the reasoning that is the "stock/default" config for Intel, but used higher speed memory for Ryzen as that is the "stock". Then they turn around and change the stock OS config and forced a non-default option. Their results were clearly flawed/wrong and yet they left that review up for 6 days before altering their review. Anytime a review clearly has an issue or not aligned with other findings, it should be removed temporarily until you can confirm your findings. Instead they chose to keep that review up, with wrong results and possibly influenced people to purchase Ryzen on false performance results. That is wrong and they deserve to be called out for making such a glaring mistake.

 

If Anandtech result show way different vs. other testers... They should move aka change how they test. Not continue in same way.

 

I have to agree with @hmscott on this one. This fix ate up the old 35% so they are now fully loaded all the time, with little room to grow. If you go to a higher sku to achieve that with Intel, the performance per $ equation skews greatly and it will shift toward AMD in some situations. That guidance is still good moving forward, but you cannot forget that it all is performance in set electrical costs and space for the budget. Now, either way, if you are not considering deployment of the first gen Epyc (meaning you already test bedded the servers and have an order, or are rerunning the Intel test beds to check before final procurement), it is a question of if you can wait for Intel's and AMD's offerings this summer. We have Cascade-SP coming, which is coffee's 14nm++ on Intel's side, then you have a potential 7nm design from AMD by the end of the year. This should have the Meltdown fix in the chip and other tweaks to help with Spectre on Intel's side, meaning your test bed and acquisitions are comparing Cascade to Epyc if starting this summer. After that, if you are waiting for next year to acquire, waiting for the new Epyc would likely give the best results. But it is all a timing thing.

 

First, you do not have to say it. The fact you didn't say anything about it IS BIAS! That is the point. If you cannot understand that at the outset, then you don't know what you are talking about and your ignorance is showing.

Also, this has been changed for a long while. Notice how you never had an issue with the numbers before now. That means that this isn't the deal you make it out to be. Now, if you are saying there should have been a disclosure, no problem. But you are not saying that. It has been a standard used by this reviewer for YEARS. It isn't new. Also, as @hmscott pointed out, RTC has been around for awhile and effected both AMD and Intel depending on CPU. Imagine that. What you have is cognitive dissonance. When both are hit by HPET, but it agrees with your bias on results, you say hooray. When it doesn't agree with your bias, even though this testing methodology has been used for much longer than releases dating to 2016, all of a sudden you complain like someone just took your child and raped your significant other in front of you. It IS NOT the armagedon you make it out to be, especially since HPET used to hit AMD CPUs harder than Intel CPUs for many years before that.

As to memory speeds, this is something AdoredTV brought up in his recent videos analyzing Stephen from Hardware Unboxed's numbers. What needs done is closer to what is being done at Gamers Nexus, where you use stock memory speeds for the baseline comparison, but then you used overclocked memory to see uplift in performance so that the consumer understands the best value and what the systems are capable of. Intel, before now, used to have higher stock memory speeds than AMD. You never complained about stock memory until now. Once again, ya' biased. If your point was that this doesn't reflect real world on the ram, then I could dig deeper into it. But you never spoke up when the stock ram speed was used by PCPer with Ryzen versus Intel, so just stop.

Onto changing settings in the OS, the OS has many moving parts. By standardizing your settings, you are actually trying to control for anomalies. If solely saying this is not informative because it doesn't reflect real use cases, that is fine, but is also true of synthetic benches to begin with. Nonetheless, that is where disclosure is proper, or doing double the testing, once at stock OS optimizations, once with HEPT on for everything, and, if you have time, with HEPT hard off both BIOS and OS. But you are not calling for that either. THAT is your problem. You are playing into tribalism rather than asking for fair disclosure and fair and full testing. Instead, you just want what you want, like a little child.

As to leaving it up, that comes to examining methodology to track down what caused the discrepancy in results. That often means messaging or calling other reviewers, running tests to confirm each element that may have caused the issues, etc. Just because you don't like the 6 days doesn't mean they were not working to correct the issue. And NO REVIEWER does that. They may, and properly should, put a note either at the top or the bottom of the article stating that the results are being challenged and a further update explaining that matter will come as soon as they have the answer. That is normal, good journalism. But pulling down a review when you have the hard data to backup your results, that is absurd. In fact, why didn't you say those things when we saw that with Ryzen 1 launch. That's right, ya biased. Seriously, you are the one that says it must ALWAYS benefit Intel, but if they do it to AMD, that's fine. So check yourself first!

 

A Timely Discovery: Examining Our AMD 2nd Gen Ryzen Results
261 comments
https://www.reddit.com/r/Amd/comments/8eucew/a_timely_discovery_examining_our_amd_2nd_gen/

dayman56 stickied comment
Ryan Smith from AnandTech said:
" Status Update: We have finished the first part of our audit of our Ryzen 2 results. The audit turned out to be a rather interesting timer issue, one that didn't impact AMD and Intel in the same way. Meanwhile, the fully updated Ryzen 2 review will be available next week."

bootgras 23 hours ago
"I can't use my 8700k system at all with HPET forced on in Windows. The entire system stutters. Had to do it when submitting some scores on hwbot. Never had that kind of experience with Ryzen"

CammKelly 20 hours ago
"First up, I come from the enterprise space, so don't take the next comments as a gamer.

We've found that in our high density hypervisors, that we get clock skew (which is a pretty big issue for us) if HPET isn't enabled. I'm not sure I like that considering the issues HPET solve, that the solution going forward is to disable HPET. It smells of bad practice.

That being said, it does appear that Intel being affected by this so hard is mostly down to Spectre\Meltdown patching, and that should be resolved by Ice Lake."