Now there's a good sign, Ryzen 2 can almost hit 6GHz

Subject: General Tech | April 17, 2018 - 02:27 PM |
Tagged: ryzen 2, overclocking, LN2, amd

It took liquid nitrogen to do, but an experienced overclocker took the 4.3GHz Ryzen 7 2700X all the way to 5.884GHz and and the 4.2GHz Ryzen 5 2600X to 5.882GHz.  If that doesn't impress you, then how about the fact that all cores were running at that speed, and not just one core active?  You will not see such high frequencies when using less esoteric cooling solutions however this indicates some serious overclocking potential for the new Ryzens in general.  Check out the proof by following The Inquirer's links here.

View Full Size

"AMD'S RYZEN 2 processors are set to be proper powerhouses as the Ryzen 7 2700X and Ryzen 5 2600X chips have been overclocked beyond a preposterously nippy 5.8GHz."

Here is some more Tech News from around the web:

Tech Talk

 

Source: The Inquirer

Video News


April 17, 2018 | 02:43 PM - Posted by unacom

That's not really saying much. Der8auer got to 5.8GHz on the 1800X without disabling cores a year ago. There's no change in this aspect.

April 18, 2018 | 08:30 AM - Posted by WhyMe (not verified)

What Der8auer achieved was probably more the exception than the rule, most first generation overclocks on LN2 Ryzen's seem to top out around the 5.3Ghz range.

April 18, 2018 | 10:16 AM - Posted by Stef (not verified)

And that could be a lucky CPU too, where did they get the CPU from? its not on sale yet so we don't know how many of them have the ability to push a single core to such frequency.

April 18, 2018 | 11:16 AM - Posted by WhyMe (not verified)

It could be but what are the chances of that, either the percentage of golden CPUs have increased massively or the average overclocks have increased by 4-500Mhz.

Occam's razor would say it's an increase in overclocking potential considering stock clock speeds have increased by 200Mhz.

April 17, 2018 | 03:21 PM - Posted by FutureMarkReMarkedToUL (not verified)

UL the owners of Futuremark is now rebranding that benchmarking software under the UL name.

"We are changing our company name from Futuremark to UL" (1)

"We're moving to a new home on April 23"

https://www.futuremark.com/pressreleases/futuremark-rebranding-to-ul-apr...

April 17, 2018 | 03:35 PM - Posted by meowster (not verified)

Go to the source and look again.
While all cores were enabled, only one core hit that high speed.

April 17, 2018 | 07:10 PM - Posted by Jeremy Hellstrom

serves me right for doing 18 things at once ... correcting and thanks!

April 17, 2018 | 04:17 PM - Posted by ThisIsJustNutzWithoutGPUTaskMonitoringAbilityInTheFingOS (not verified)

Now Intel(1) has found a use for its Integrated Dogfood Graphics for malware scanning acceleration usage.

OH if those D--k Heads in Redmond would just update Task Manager to show any task/s running on the GPU. And Intel is out of its collective mind if they think that anti-malware engines can not be hijacked to do nefarious things via any processor CPU, GPU, otherwise.

This can not end well for some users of M$'s security software/other's software that may want to take advantage of any Intel Framework/API that allows GPUs to scan system memory for malware.

And What about those Lords of Redmond even devoting the necessary resources to allow Windows Task Manager to track any task sent to any GPU/s under that Windows OS Known as 10 and all that all spyware baked in.

How many decades did in take for the Folks in Redmond to even add the ability of Task Manager to even track GPU usage and now there is still a need for Task Manager to track GPU tasks to this very day and that's still not offered even for Windows 10. I'm sure not trusting Intel or M$ and GPUs made available for other workload acceleration really should indicat to M$ that maybe their "OS" should have the ability to track any tasks sent to the GPU especially tasks that make use of the GPU to scan system memory!

(1)

"Intel's security light bulb moment: Chips to recruit GPUs to scan memory for software nasties"

https://www.theregister.co.uk/2018/04/17/intel_gpu_malware_detection_sec...

April 17, 2018 | 05:37 PM - Posted by MarktingLardzAndRancidLipids (not verified)

Woopz Ninjaed by that other article listed, but srill the Reg commentards so love to tell is like it is about the Crony Capitalist Monopolies that are so in control in the States. It's so refreshing to see compared to the professional career service yes men brown noses here in the Pyramid Scheme Of Scammerica.

April 17, 2018 | 04:45 PM - Posted by CK (not verified)

Maybe it is a dumb question but what does the SEP abbreviation stand for? The one after the price for the coming Ryzen+ parts.

April 17, 2018 | 05:29 PM - Posted by MarktingLardzAndRancidLipids (not verified)

SEP(Suggested Etailer Pricing)

Some name for online retailer pricing that's really F-ed up that the marketing F--ks have thought up.

And it's still hard to find AMD Raven Ridge based APU's in most brick and mortar outlets in resonable offerings.

April 17, 2018 | 07:31 PM - Posted by CK (not verified)

Thanks

That is weird they usually use the MSRP (manufacturer's suggested retail price) abbreviation.

April 18, 2018 | 04:10 PM - Posted by James

It is kind of surprising that they are getting this high of clock speeds. They claim to be using a more GPU like process tech rather than a higher performance (but also higher power consumption and lower density) process normally used for CPUs.

I was just looking at the ISSCC Zeppelin slides. The Zeppelin die actually has an extra on package IF link that I hadn’t noticed before. It actually has 4 IFOP (Infinity Fabric On Package) in addition to the two IFIS (Infinity Fabric Inter-Socket) links. Only one of the two IFOP on the DDR interface side of the die is actually used, but which one depends on chip placement. The two IFIS links are IO, except in a dual socket system where one is used to route to a die on the other socket. So AMD actually included an extra x16 IFOP link just to make the package routing simpler and cheaper; it can be done in only 4 pcb layers. It is interesting that this is a worthwhile trade off between die size and package cost. I expect that the IFOP links take significantly less die area than an IFIS link though. Routing distance is only a few centimeters at most.

The clock speeds they are getting seem to indicate that using the higher density process and design libraries is a worthwhile trade off also. Current CPUs are heat limited, so you generally aren’t going to run at maximum boost clock for very long. Most of the things I care about are long enough running that it causes most CPUs to drop to their base clock anyway.

April 18, 2018 | 04:28 PM - Posted by seika (not verified)

so what if they can be clocked that high? what's the point if the ipc isn't as high as intel's offerings. i'm pretty sure a locked intel i7-8700 would smoke any ryzen 2 chip in gaming.

April 19, 2018 | 01:12 AM - Posted by James

More cores for the same or lower price is the usual offer. The IPC should be very close for this round. Intel hasn’t really improved IPC much in a while. Intel has been holding the industry back for a long time by keeping the mainstream at 4 cores. Games can use 4 cores relatively easily, but scaling up to many cores takes a bit of software work. Intel didn’t want consumer software to be optimized for more than 4 cores because AMD was offering 8 core processors. The excavator processors weren’t great at single threaded applications, but it still would have gotten a big boost from properly multi-threaded software. So instead, everyone got 4 cores with integrated graphics where another 4 cores could have gone. This also allowed Intel to price gouge on workstation/server parts that could use more cores. 8 core processors should have been mainstream at 22 nm.

I have some 24 core machines at work (dual 12 cores with HT off), and it is really pitiful to run a single threaded application that only uses 4.17% of the available cpu power. We already had issues becaus someone spec’ed the machines with a lower number of cores at a higher clock speed. We had a lot of trouble reaching performance targets. A few hundred MHz is nothing compared to having many more cores when the software is well threaded. Four cores is a bit of a boundary; games can easily take advantage of 4 cores without really doing much multithreading work due to just the nature of the code. Once the work is done in game engines to really take advantage of many cores, then a little higher clock or a little higher IPC isn’t going to compete. More cores will win by a large margin in the long run. We need DX12 and Vulkan to make that happen though, and it has been in Nvidia’s best interest to hold that back since that would also help AMD. We are getting to the point where Nvidia is going to be seriously limited by DX11 also. It has a major bottleneck that all draw falls have to be submitted on a single thread. Recombining draw calls after distributing them out to multiple threads is a bad design. DX12 and Vulcan can have multiple threads submitting work to the GPU. It will scale significantly better.

While I am a bit of an enthusiast myself, the enthusiast attitude that everyone should pay a large amount extra just to get a few percent higher performance is ridiculous and it is how we got to the situation with Nvidia and Intel having near monopolies. Those near monopolies are why we have had essentially stagnation for the last several years. Several of my friends have 6 and 8 core AMD excavator parts and they have been perfectly happy with them. They just do general compute stuff like web browsing and maybe playing World of Warcraft. That doesn’t take much cpu; they got an AMD 390 video card which is probably over powered for what they do. Enthusiast always say they should have gone intel, but it would have been more expensive, and with better threaded software now, those 6 and 8 core AMD excavator parts probably outperform the 2 and 4 core Intel competition of the time.

May 27, 2018 | 08:36 PM - Posted by drwho (not verified)

i have cpuid hwmonitor screen shot, but it doesn't look like i can post images. i did not intend to max out my cpu's capabilities but i like to monitor cpu temp. after playing overwatch for ~2hr, cpuid-hwm showed peaks of 6.035ghz on one core and 5.963ghz on two others.

ryzen 2600x
coolermaster 212 led cpu cooler
collermaster 912 haf case + two 200mm fans (front, top)
gigabyte aorus 5 gaming mb
32gb corsair vengence 3200 ram
evga gtx 980
evga g3 750 power

6ghz on 3 cores @ 88'C
no cores disabled.
ram, gpu, cpu at default settings

July 12, 2018 | 09:58 AM - Posted by scottoknow (not verified)

I have a 2600x. After having some fun on oculus. CPUID HM stated that all cores where running at 4.7ghz. Can that be right?.

July 12, 2018 | 09:58 AM - Posted by scottoknow (not verified)

I have a 2600x. After having some fun on oculus. CPUID HM stated that all cores where running at 4.7ghz. Can that be right?.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.