Jump to content
stevelees

32 Windows Xeon Cores Slower than 8 Core i7 Macbook Pro

Recommended Posts

I recently dumped Mac (and Adobe) for Windows/Affinity and spent big money on a HP Z840 with 2 x Xeon 2.1GHz processors each with 8 cores and 16 multi-threaded cores, (totalling 32 CPU cores running as 2.1GHz) 32GB RAM, Nvidea GTX 1080i Video Card, running on an ultra fast HP SSD (3000MB/s read speed/2000MB/s write speed).

 

A friend came over today and we compared speed of Affinity Photo opening a very large PSD file, and saving a JPEG of that file. He had his 4 year old Macbook Pro running an old 4 core i7 at 2.6GHz, with 8 multi-threaded cores.

 

The result was embarrassing.....His old laptop opened the 8GB file much quicker than my machine which has more cores and a way faster SSD. Saving the JPEG file was also much faster on the Mac.

 

What I noticed was that the windows machine uses only a single core for the opening of a file, and only one core for the saving of the file, whereas the Mac used four cores. Why?

 

We then compared the speed when doing a gausian blur on the file, we started both at the same time and all 32 cores on my Windows machine ran at 100% and the progress bar flew to the end and sat there doing seemingly nothing, (just when I thought it was going to annihilate the Mac) but I noticed it then switched back to using a single core to do what exactly I do not know, but it allowed the old Mac to catch up, making the time difference minimal. The Mac was using all 8 cores for this process.

 

So, my question is, WHY is the Windows version seemingly limited to single core for certain tasks, making it's performance hobbled on a machine like the Z840? My preference for ANYTHING I do is to have the app I am working in making full use of all resources on my machine and Affinity is not doing that at all times, yet you allow it on Mac, makes no sense....what am I missing?

 

Share this post


Link to post
Share on other sites

Very specific interesting questions, topic followed!

 

What he said! (Except that I won't bother adding it to my 'Content I follow', now that I've posted to the thread and can find it that way.)


Alfred online2long.gif
Affinity Designer/Photo/Publisher for Windows • Windows 10 Home (4th gen Core i3 CPU)
Affinity Photo for iPad 1.8.6 • Designer for iPad 1.8.6 • iPadOS 14.3 (iPad Air 2)

Share this post


Link to post
Share on other sites

Hi stevelees,

Welcome to Affinity Forums :)

Only an Affinity dev can reply to your question, but overall, the Windows version was launched just six months ago while the Mac version already has more than two years of development. We are still stabilising/fixing bugs that were reported after the Affinity's launch on Windows (mostly due to the huge variation in software/hardware configs) so i'm not surprised if in some aspects the Windows version isn't performing at its best. Hopefully the gap will disappear as we move forward with the development.

Share this post


Link to post
Share on other sites

Hi Meb,

 

Thank you for the welcome!

 

I guess that makes some sense given the recent birth of the Windows version. I ran the test that was listed here:

 

https://forum.affinity.serif.com/index.php?/topic/32907-affinity-photo-performance-comparison-data-sheet/?p=198410

 

It would seem that overall the performance I am getting is pretty impressive on that test, but I think real world daily functions such as file opening and JPEG exporting would have been the first things addressed rather than an after thought.

 

If I can make a suggestion, it would be great to have a PREFERENCE setting in the performance part to choose how much of the computers resources are to be assigned to the Affinity Photo, but then again it does grab everything the machine has to offer for certain tasks, so it is bewildering that the basic functions are so slow. 

 

Anyway, I hope it is addressed soon as I have removed Adobe from my life and have no regrets, I just want to see Affinity go from strength to strength and fixing things like this would go a long way towards it I think.

Share this post


Link to post
Share on other sites

Well, you are the speed champion on that test so far. Which is good. :)

 

I hope your experience will put on the table issues that were not noted before so that the Dev Team can do somethign about it.

 

And welcome aboard!


You'll never know what you can do until you get it up as high as you can go!   

AMD FX 8350 :: Radeon HD 7870 :: Windows 10 ::  http://mithferion.deviantart.com/

Oxygen Icons :: Free Quality Fonts :: Public Domain Pictures :: iOS 11 Design Resources :: iOS App Icon Template :: Hot to do High Quality Art :: Mesh Warp / Distort Tool Considerations

Share this post


Link to post
Share on other sites

Well, you are the speed champion on that test so far. Which is good. :)

 

I hope your experience will put on the table issues that were not noted before so that the Dev Team can do somethign about it.

 

And welcome aboard!

 

Cheers mate thank you.

Share this post


Link to post
Share on other sites

...

 

A friend came over today and we compared speed of Affinity Photo opening a very large PSD file, and saving a JPEG of that file. He had his 4 year old Macbook Pro running an old 4 core i7 at 2.6GHz, with 8 multi-threaded cores.

 

The result was embarrassing.....His old laptop opened the 8GB file much quicker than my machine which has more cores and a way faster SSD. Saving the JPEG file was also much faster on the Mac.

 

What I noticed was that the windows machine uses only a single core for the opening of a file, and only one core for the saving of the file, whereas the Mac used four cores. Why?

 

We then compared the speed when doing a gausian blur on the file, we started both at the same time and all 32 cores on my Windows machine ran at 100% and the progress bar flew to the end and sat there doing seemingly nothing, (just when I thought it was going to annihilate the Mac) but I noticed it then switched back to using a single core to do what exactly I do not know, but it allowed the old Mac to catch up, making the time difference minimal. The Mac was using all 8 cores for this process.

 

So, my question is, WHY is the Windows version seemingly limited to single core for certain tasks, making it's performance hobbled on a machine like the Z840? ...

 

Well usually what ever processes AP throws on it should fly here, according to your used hardware it should be light years ahead under Win instead in contrast to that Macbook Pro, even I/O processes. But and that's a big but, I believe the AP Win port was made somehow the quick, under time to release pressure way here, so they didn't found time to adapt and compile the whole code in a more Win CPU/GPU optimized way. - However, hope dies last, therefor let's see what the future brings and if the Win versions will get the needed speed push ups and stability in future versions, when bottlenecks have been sorted out and optimized!


☛ Affinity Designer 1.8.4 ◆ Affinity Photo 1.8.4 ◆ OSX El Capitan

Share this post


Link to post
Share on other sites

I'm heartened by MEB's comments. Reading between the lines I'd guess the devs are well aware of where the bottlenecks are but more importantly why. They have their priorities, stabilising/fixing bugs (the tablet issues are probably taking up much more time and resource than planned for) but performance would seem to be next.

It's easy to forget this software is only six months old, I feel like I've been using it for years! good reminder MEB

Share this post


Link to post
Share on other sites

I'm heartened by MEB's comments. Reading between the lines I'd guess the devs are well aware of where the bottlenecks are but more importantly why.

Don't overlook the importance of MEB's comment about the huge variation in the number of hardware/software configurations the Windows versions have to support. On the Mac side, there are at most a couple of dozen different hardware configurations they have to support, & most of them are limited to the old "cheese grater" Mac Pros that can use a limited number of non-OEM graphic cards, which still use the same or very similar software drivers as the OEM ones.

 

On the Windows side, there are hundreds if not thousands of different hardware configurations to support, plus all the different driver softwares that might be installed from various sources. So in effect, there could be hundreds of bottlenecks, each affecting some but not other Windows systems. It seems likely that it will not be a quick or easy task to find & eliminate all of them!


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

I've been reporting about issues with dual Xeon for quite a while now...

And to whoever think the GPU makes a difference, no, it doesn't. Tested on a double Titan x.

It's a tough one, really hope they can get this sorted out.


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

The bottleneck in this case is rather simple: The Mac uses all CPU cores when opening a file, and exporting a JPEG. The Windows version uses a single core for both tasks. To me at least that has nothing to do with the number of variations they need to contend with in the Windows world, but rather a simple command they have set to use one only. I say this as it clearly uses all 32 cores when running the test here:

 

https://forum.affinity.serif.com/index.php?/topic/32907-affinity-photo-performance-comparison-data-sheet/?p=198410

 

So why allow ALL CPU cores for many of the tasks in the above test, and limit it to one on the tasks you do every time you use AP? (ie: Opening a file!)

Share this post


Link to post
Share on other sites

And to whoever think the GPU makes a difference, no, it doesn't.

Why do you say that? There is more to it than just which GPU is used, or how many cores are used for some particular thread. Just using a different graphics driver than what the software was designed to use can have a significant effect, as can anything else that affects any runtime services provided by the OS.


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

Why do you say that? There is more to it than just which GPU is used, or how many cores are used for some particular thread. Just using a different graphics driver than what the software was designed to use can have a significant effect, as can anything else that affects any runtime services provided by the OS.

 

I say that because GPU is another element mentioned in other threads as an important part in order to get better performance.

I don't know whether it is true or not, not sure how much Affinity takes advantage of GPUs (I would expect it does), but I can definitely tell that even two Titan x didn't make any difference.

I know how much a graphic driver can affect the performance of a program, I had to deal with 3ds max for years. When it comes to Affinity, though, it's kind of pointless as there is no official statement of what driver is optimized.


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

When it comes to Affinity, though, it's kind of pointless as there is no official statement of what driver is optimized.

If there are dozens to hundreds of driver/GPU combinations to consider, how practical would it be for them to try to do this? How could they even present all that data in a usable format?


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

If there are dozens to hundreds of driver/GPU combinations to consider, how practical would it be for them to try to do this? How could they even present all that data in a usable format?

 

R-C-R,

 

I'm not sure we are talking about the same thing here.

When a software house optimizes its product for a driver, then they can clearly tell what driver they recommend.

That's, for example, what Autodesk has been doing for years.

Obviously, Affinity is not yet optimized for any specific driver or even video card otherwise we would read a clear statement in that regard (in fact, last time I checked there was no specific card nor driver).

Hence, my consideration in the post above: talking about drivers at the moment is pointless and, for the same reason, to whoever thinks GPU is what actually matters in order to make Affinity run faster, it is not, at least not yet.


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

If there are dozens to hundreds of driver/GPU combinations to consider, how practical would it be for them to try to do this? How could they even present all that data in a usable format?

 

Well that's nonsense, when programming graphics related things you don't reinvent the wheel here, you don't take any single possible driver into account. Instead you use abstraction layers between algorithmns and the hardware equipment of computers, that's what those APIs like MS AMP (C++ AMP Accelerated Massive Parallelism) etc. are for. - So the common way this works is, if for example an AMP-using program does not provide a sufficiently up-to-date GPU, the algorithm programmed using AMP is automatically executed on the CPU using its parallelization options (multithreading on multiple processor cores, SIMD instructions). AMP is thus intended to provide an abstraction layer here between an algorithm and the hardware equipment of the executing computer.


☛ Affinity Designer 1.8.4 ◆ Affinity Photo 1.8.4 ◆ OSX El Capitan

Share this post


Link to post
Share on other sites

Well that's nonsense, when programming graphics related things you don't reinvent the wheel here, you don't take any single possible driver into account.

But that seems to be essentially what verysame wants: some kind of clear info about what driver and/or graphics card Serif recommends.


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

But that seems to be essentially what verysame wants: some kind of clear info about what driver and/or graphics card Serif recommends.

 

R-C-R,

 

I'm don't personally want anything, I was pointing out that a good GPU is not a guarantee of a smooth experience with Affinity.

I was adding that what software houses do, is releasing a list of certified hardware and since Affinity hasn't so far, that's one more reason to not expect any particular performance improvement because of this or that GPU. Again, if such specs list for Affinity exists, I would love to see it as I might have overlooked it.

 

I don't know about the technicality in order to come up with such a list, but is nothing new.

I'm surprised we have spent this back and forth on this matter, I assumed you were aware of such specs companies publish.

 

Anyway, this may give you some hints:

http://usa.autodesk.com/adsk/servlet/syscert?siteID=123112&id=18844534&results=1&stype=graphic&product_group=6&release=2015&os=8192&manuf=all&opt=2

https://helpx.adobe.com/after-effects/system-requirements.html

 

On top of that, usually when a new video card come out what these companies do is releasing an update for their product as well in order to add support to that particular card. The way you put it instead, sounds as if due to the amount of possible combinations, this task is impossible to tackle. Well, that is simply not true and incorrect, otherwise how AD and Adobe would do it?

 

Hope this will help you to understand what I tried to express in my previous posts.

 

Best,

Andrew


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

verysame,

 

I am aware of the kind of hardware lists some companies publish as "certified" or "recommended." However, note for instance the disclaimer at the bottom of the Autodesk page (emphasis added):

Autodesk certification testing focuses on workstation-grade hardware, as identified by the hardware vendor. Testing and designation as Certified or Recommended does not guarantee that a particular device is workstation-grade or that it will meet a user's particular needs. Test results are valid only for the tested combination of hardware and driver. Certified or Recommended status does not guarantee that the hardware will perform acceptably with other drivers.

 

Basically, all they are saying is the listed cards passed the tests they performed listed on the linked pages when used with the specific driver shown on that page & the indicated OS version. In this context "passed" just means the card & driver configuration did not fail to complete the test, & completed it without issues. It does not say anything about how well the configuration performed the listed tests.

 

Likewise, Adobe's list just shows cards their apps support for various functions, but that does not say anything about how well the listed cards perform.

 

True, Serif doesn't publish such lists for the Affinity line but since these lists really just indicate what cards/drivers are supported, instead they do what most companies do, which is to list a generalized minimum specification that must be met, in this case "DirectX 10-compatible Graphics Cards and above."

 

Optimizing an app's performance for a particular hardware/software configuration is an entirely different thing from supporting it, & that is a much more involved process, requiring not just pass/fail testing but also tweaking the code, providing optimized exception handlers, & so on.


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

First off, you are basically confirming what I've been saying, that is when people on this forum claims that GPU are important in order to get better performance from Affinity's, that statement is not backed up.

 

Nonetheless, if a video card is claimed to be supported by an application that relies on GPU calculations, and since we know that GPU renders are marketed only for one thing when it comes to graphics applications, which is speed, then you can easily do the math.

 

Last but not least, if you are a 3ds Max user you know what that AD's statement mean, otherwise you are nitpicking around terminology.

When the hardware is certified by AD, that means that the user will benefit from stability and speed, which is why hardware acceleration comes into play with certified hardware.

At the moment, no one knows what hardware/driver can be efficient for Affinity products, not even the devs (otherwise, they will tell us).

 

But then again, to your point:

Certified or Recommended status does not guarantee that the hardware will perform acceptably with other drivers.

 

Which means exactly what we've been discussing so far: the fact that only what's on that page (hardware and driver) is certified. Any other combo, it's at user's risk.

 

And by the way, tests are meant to check in both areas, stability and performance.

A pretty common one is Cinebench from Maxon: https://www.maxon.net/en-us/products/cinebench/

 


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

verysame,

 

Just to be as clear as I can be about this, earlier you wrote (emphasis added):

When a software house optimizes its product for a driver, then they can clearly tell what driver they recommend.
That's, for example, what Autodesk has been doing for years.

 

That is not what Autodesk has been doing. Like I said, supporting is not the same thing as optimizing.

 

In the same post you said, "talking about drivers at the moment is pointless," & yet both of us agree that they can make a difference in performance.

 

So, to get back to what I was saying earlier, because there are so many more software & hardware possibilities on the Windows side than on the Mac side, optimizing Affinity for the various Windows configurations is going to take much longer than it has for the Macs. It would be helpful if, in addition to reporting basic hardware info like CPU & GPU specs, Windows users would also report what driver(s) & version(s) are installed, & anything else that could help differentiate their configurations from others, be it in the hardware, the firmware, or the system or application level software.

 

It is unrealistic to expect the Affinity team to test all of these configurations, much less to optimize the apps for them all, but info like this can help them determine where to look to make improvements.


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

verysame,

 

Just to be as clear as I can be about this, earlier you wrote (emphasis added):

That is not what Autodesk has been doing. Like I said, supporting is not the same thing as optimizing.

 

In the same post you said, "talking about drivers at the moment is pointless," & yet both of us agree that they can make a difference in performance.

 

So, to get back to what I was saying earlier, because there are so many more software & hardware possibilities on the Windows side than on the Mac side, optimizing Affinity for the various Windows configurations is going to take much longer than it has for the Macs. It would be helpful if, in addition to reporting basic hardware info like CPU & GPU specs, Windows users would also report what driver(s) & version(s) are installed, & anything else that could help differentiate their configurations from others, be it in the hardware, the firmware, or the system or application level software.

 

It is unrealistic to expect the Affinity team to test all of these configurations, much less to optimize the apps for them all, but info like this can help them determine where to look to make improvements.

 

Perhaps a bad choice of words from my side, but if a product runs better on a device rather than another, it clearly is optimized for that particular device.

 

That said, I don't find it fair when people quotes others' by only reporting part of what they said, as the result can sounds quite different.

If you wouldn't often bring technical topics to the table, I would say you were a political figure in the past.

This is what I said:

 

[...] it's kind of pointless as there is no official statement of what driver is optimized.

 

So, yes it can makes a difference, but as of now what's the purpose of all this talking as no one, not even Serif's devs can tell which hardware configuration works better?


Andrew
-
Win10 x64 AMD Threadripper 1950x, 64GB, 512GB M.2 PCIe NVMe SSD + 2TB, dual GTX 1080ti
Dual Monitor Dell Ultra HD 4k P2715Q 27-Inch

Share this post


Link to post
Share on other sites

I do not know what your purpose is but mine is to encourage Windows users to report more than the basic hardware info. I have made my reasons for that as clear as I can.


Affinity Photo 1.8.4, Affinity Designer 1.8.4, Affinity Publisher 1.8.4;  2020 iMac 27"; 3.8GHz i7, Radeon Pro 5700, 40GB RAM; macOS 10.15.6
Affinity Photo 
1.8.4.186 & Affinity Designer 1.8.4.4 for iPad; 6th Generation iPad 32 GB; Apple Pencil; iPadOS 14.0.1

Share this post


Link to post
Share on other sites

I do not know what your purpose is but mine is to encourage Windows users to report more than the basic hardware info. I have made my reasons for that as clear as I can.

 

That's when tets like the one with the Macros help, I guess.

 

Best regards.


You'll never know what you can do until you get it up as high as you can go!   

AMD FX 8350 :: Radeon HD 7870 :: Windows 10 ::  http://mithferion.deviantart.com/

Oxygen Icons :: Free Quality Fonts :: Public Domain Pictures :: iOS 11 Design Resources :: iOS App Icon Template :: Hot to do High Quality Art :: Mesh Warp / Distort Tool Considerations

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

×
×
  • Create New...

Important Information

Please note the Annual Company Closure section in the Terms of Use. These are the Terms of Use you will be asked to agree to if you join the forum. | Privacy Policy | Guidelines | We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue.