Tuesday, April 4th 2017

Outertech adds AMD Ryzen Processor Optimization to Cacheman 10.10

Outertech has released Cacheman 10.10, a Windows performance enhancement software that uses one-click optimization in order to improve responsiveness, privacy, and the security of a PC. The new Cacheman version introduces support for the recently released AMD Ryzen 7 1700, 1700X, 1800X and Ryzen 5 1400, 1500X, 1600, 1600X processors. A free test version is available from the Outertech website.

The AMD Ryzen 7 processors consist of 16 CPU cores - 8 physical and 8 virtual (emulated) cores. The physical processor cores are placed on the CPU die in two groups of 4 cores each, the so called CCX (CPU Complex). The two groups are interconnected with a 256-bit wide bi-directional crossbar. The speed of the crossbar is linked to the speed of the system memory. Within a CCX group, CPU cores can communicate very quickly with each other. Communication between cores that sit on separate CCX groups is significantly slower (by the factor of 2 and more).
DOWNLOAD: Outertech Cacheman 10.10

Cacheman (short for Cache manager) auto-optimizes the Windows sub-systems including cache and memory in order to improve responsiveness, privacy, and security. In addition to the already built-in Intel Core processor optimization, Cacheman 10.10 updates the optimization profiles for AMD Ryzen 7 1700, 1700X, 1800X and Ryzen 5 1400, 1500X, 1600, 1600X processors. A range of computer performance profiles is available including gaming, graphics workstation, digital audio recording, notebook, and server.

After a discovery made in the testing lab Outertech recommends an additional manual optimization that can be performed within the Cacheman user interface.

Outertech has discovered that tying some Windows applications to the first Ryzen 7 CPU CCX group (4 physical + 4 virtual processor cores) can increase the performance by a significant factor, as thread switching between two CCX groups is avoided. This will work well only with applications that do not make full use of all 16 CPU cores, particularly computer games. Windows 10 appears to not be aware that the Ryzen processor consists of two individual CPU core groups. Switching program threads from one CCX group to another can cause performance degradation on an otherwise very fast processor. Outertech expects a fix from Microsoft on this issue and has decided against an automatic Cacheman optimization of this particular issue at this time. Users can manually try this optimization by right clicking the name of the application in the Cacheman user interface and selecting Set Core Affinity (sticky) | Core 1-8 from the menu. Cacheman will remember this tweak and apply it automatically each time the game is started.

Cacheman starts at USD 28.95.
Add your own comment

25 Comments on Outertech adds AMD Ryzen Processor Optimization to Cacheman 10.10

#1
bogami
Onle for 28.95 USD !
Posted on Reply
#2
the54thvoid
Hmm... You'd think AMD would release such a software fix if it was indeed such an issue...
Posted on Reply
#3
RejZoR
Wait a sec, didn't they say Windows scheduler works correctly with Ryzen. And now they are saying Windows is not CCX aware. Which one is it now?
Posted on Reply
#4
dj-electric
Who here has Ryzen and wanna do some testing for us?
Posted on Reply
#5
_JP_
btarunr
Users can manually try this optimization by right clicking the name of the application in the Cacheman user interface and selecting Set Core Affinity (sticky) | Core 1-8 from the menu. Cacheman will remember this tweak and apply it automatically each time the game is started.
Or...you know, just create a shortcut that uses that start command with /affinity.
Same concept and one less piece of software installed -> Optimization :)

EDIT: Added quote
Posted on Reply
#6
Mysteoa
RejZoR
Wait a sec, didn't they say Windows scheduler works correctly with Ryzen. And now they are saying Windows is not CCX aware. Which one is it now?
I think it was combination of things. That Windows is not CCX and Hyperthread aware, but it was proven that it was not true for the Hyperthread part. It is possible to further tune the scheduler for the CCX
Posted on Reply
#7
HD64G
Windows need to be fixed for Ryzen to perform optimally for all the sw needs. That's for sure due to Ryzen's new arch being totally different from all previous CPUs. Some updates are rumoured to be included in latest patches from MS. Maybe in @W1zzard 's test coming later we shall see some changes in performance from newer BIOS, faster RAM and OS updates.
Posted on Reply
#9
the54thvoid
Dj-ElectriC
Who here has Ryzen and wanna do some testing for us?
Send me thirty dollahs and I make you results for cheap.
Posted on Reply
#10
punani
Dj-ElectriC
Who here has Ryzen and wanna do some testing for us?
@W1zzard :respect:
Posted on Reply
#11
VSG
Editor, Reviews & News
There's a free trial version available. The whole website reminds me of a free/shareware one from 2005 though.
Posted on Reply
#12
Mirkoskji
System explorer does the same for free. Here you pay for some sort of auto optimization
Posted on Reply
#13
danaj525
Dj-ElectriC
Who here has Ryzen and wanna do some testing for us?
Depends what you want tested? Let me know and I'll see what I can do!
Posted on Reply
#15
Foobario
As has been demonstrated by those that have gotten their DRAM up to 3200, the latency becomes much less of an issue as the "bus speed" between the two four core blocks speeds up as well. So BIOS updates that make 3200 ghz the norm will up Ryzen's performance beyond it's more than capable levels now, for now.

Optimizations that keep the workload within the same four cores and staying in sync with it's assigned cache will move performance of Ryzen even further. I'm betting that it will make memory speed less important in the future.

Speaking of optimizations, yes, AMD said all was fine with Microsoft's distribution of workloads. I suspect that was to avoid embarrassing Microsoft or projecting blame on them. AMD may have rushed Zen a bit, as some have suggested, and didn't give Microsoft time to get real optimizations in place or Microsoft just dragged their asses on it. Most likely a combination of both.


AMD is in the running for some of the Surface 5 production. Makes sense that AMD would go out of their way to avoid blaming Microsoft for anything.

Regardless, Microsoft will create Ryzen optimizations and release them, possibly, with the Surface 5 launch. Or without giving AMD any Surface 5 business.

A month or two from now Ryzen will be performing better than it is now for anybody on Windows 10. If you can't wait, this Cacheman 10 product might make you the cool kid on the block, for only $30, until everybody on Ryzen gets the free Windows 10 upgrade.
Posted on Reply
#16
Caring1
RejZoR
Wait a sec, didn't they say Windows scheduler works correctly with Ryzen. And now they are saying Windows is not CCX aware. Which one is it now?
By stating the scheduler works fine, AMD are admitting in a round a bout way their design is not as efficient as it should be.
Posted on Reply
#17
medi01
Fluffmeister
CCX seems to be a pain in the ass.
No, CCX seems to be a corner pragmatically cut, especially given AMD's budget.
4+0 is within 5% of 2+2

Posted on Reply
#18
RejZoR
Caring1
By stating the scheduler works fine, AMD are admitting in a round a bout way their design is not as efficient as it should be.
Well, pulling what they did here with the budget they had/have, the fact that CCX isn't "perfect" design is irrelevant imo. The fact they can offer such exceptional CPU's for such price makes this little "issue" irrelevant in my books. And many people share the same opinion.
Posted on Reply
#19
Khanivore
Is Process Lasso similar to Cacheman?
Posted on Reply
#20
bug
Fluffmeister
CCX seems to be a pain in the ass.
Honestly, I doubt it matter much on the desktop. On servers, that's where it can hurt.

CCX is what Intel did with Pentium-D. Only this time the two dies talk over a slow connection. The impact is data available in L3 cache of the other CCX is accessed at the same speed as if it wasn't cached at all. On the desktop, the user is the limiting factor most of the time anyway. But servers and machines that do number crunching may see a whole different story altogether.

Since I'm neither building a server nor upgrading my desktop, I can enjoy my popcorn on the sideline.
Posted on Reply
#21
Foobario
Had AMD not taken the CCX approach they would have been forced to price the 1800X at $800, or more, instead of $500 due to the much worse yields an 8 core chip has compared to a 4 core chip.

By taking this approach, AMD is basically producing only four core dice and nothing else which will allow GLOFO to become extremely efficient at such production much quicker.

The offset to this more cost effective production method is having to do extra work with the software industry to allow them to optimize for the partitioned blocks.

Ironically, the optimizations that are being done for Ryzen today will give Intel an easier introduction of their partitioned 8 core chips in the future. As they will have to bring their production costs down if they want to be competitive in the price/performance ratio in the 8 core market.
Posted on Reply
#22
efikkan
Aaah, optimizer software. It's sad to see this kind of crapware still gets any attention.

Just the other day I watched an old episode of The Computer Chronicles, featuring in part such software. I remember "optimizers" being a big thing in the DOS era, and even in the late 90s on Windows 9x. It was a really thriving market back in those days when hardware were expensive, even though most of them appeared to be crap slapped together in VB which did little or nothing good.

The problem with "optimizers" is that there is really little to optimize in an operating system this way. Granted, setting manual affinity can help in a few edge cases, and ramdisks have their use cases, but otherwise they are mostly doing nothing at all or inflicting actual harm. Some utilities tries to free memory, which means they will free shared memory, which may result in undefined behavior/crashes for programs. Registry cleaners can be really harmful. Most of these utilities should be considered as malware.
Posted on Reply
#23
WaroDaBeast
FreedomEclipse
¸.•*¨*•♫♪¸¸.•*¨*•♫♪ Its the Cacheman!¸.•*¨*•♫♪¸¸.•*¨*•♫♪ Ski-bi dibby dib yo da dub dub¸.•*¨*•♫♪¸¸.•*¨*•♫♪


I'll go get my coat....
Don't you move an inch!!! Let's sing together! ¸.•*¨*•♫♪¸¸.•*¨*•♫♪ Ba-dop-ba-dop, BEEEE-bop-bop-bop, bop-dop-bop! ¸.•*¨*•♫♪¸¸.•*¨*•♫♪

Damn, I love this song's powerful message.
efikkan
Aaah, optimizer software. It's sad to see this kind of crapware still gets any attention.

Just the other day I watched an old episode of The Computer Chronicles, featuring in part such software. I remember "optimizers" being a big thing in the DOS era, and even in the late 90s on Windows 9x. It was a really thriving market back in those days when hardware were expensive, even though most of them appeared to be crap slapped together in VB which did little or nothing good.

The problem with "optimizers" is that there is really little to optimize in an operating system this way. Granted, setting manual affinity can help in a few edge cases, and ramdisks have their use cases, but otherwise they are mostly doing nothing at all or inflicting actual harm. Some utilities tries to free memory, which means they will free shared memory, which may result in undefined behavior/crashes for programs. Registry cleaners can be really harmful. Most of these utilities should be considered as malware.
I think optimizer software was, to an extent, useful under Windows 98 (and probably 95, but I haven't used in enough to actually know). There are so many things that OS failed to do (F***ing. End. That. Task!) or just plain didn't have (RAM defragmenting) that some tools were useful. Oh, and registry cleaners. There were so many useless things left over after uninstalling apps. I've always been kind of a diligent person, so I'd go and clean stuff from the "Program Files" folder and from the registry, but God, that was a hassle.
Posted on Reply
#24
efikkan
WaroDaBeast
I think optimizer software was, to an extent, useful under Windows 98 (and probably 95, but I haven't used in enough to actually know). There are so many things that OS failed to do (F***ing. End. That. Task!) or just plain didn't have (RAM defragmenting) that some tools were useful. Oh, and registry cleaners. There were so many useless things left over after uninstalling apps. I've always been kind of a diligent person, so I'd go and clean stuff from the "Program Files" folder and from the registry, but God, that was a hassle.
I hope you don't think "RAM defrag" can help, because the internal allocations inside a memory page is unknown even to the kernel. In theory, each page might contain as little as a single allocated byte, but only the program itself will know that. Any attempt to rearrange this data into fewer pages will end in tears. And with RAM being random access and all, aligning the pages wouldn't matter either, and remember that the program internally doesn't use hardware addresses, but virtual addresses.
Posted on Reply
#25
WaroDaBeast
efikkan
I hope you don't think "RAM defrag" can help, because the internal allocations inside a memory page is unknown even to the kernel. In theory, each page might contain as little as a single allocated byte, but only the program itself will know that. Any attempt to rearrange this data into fewer pages will end in tears. And with RAM being random access and all, aligning the pages wouldn't matter either, and remember that the program internally doesn't use hardware addresses, but virtual addresses.
I stand corrected, good sir.
Posted on Reply
Add your own comment