[OFFICIAL] Windows 10 (Build 10.0.10240) discussion & upgrade experience

[adrynalyne]

So it's not true that it's going to yield gains across the board?

 

On a related note, this is what 3D mark told me 

 

Also, I was reading this and it interested me. Would this make the GPU on my processor worth turning on again, or would I have to wait for Intel and AMD to enable cross GPU support first?

So where is it 10% slower, or is it a different test?

 

It seems to me that performance increases across the board still require api recognition.

 

For example, a DX11 card will perform better with a DX12 enabled app than with just DX11, but not as good as a DX12 card.

[Noir Angel]

So where is it 10% slower?

 

This was a test ran with the latest version of 3D mark, it was slower in an older version of 3D mark.

[adrynalyne]

This was a test ran with the latest version of 3D mark, it was slower in an older version of 3D mark.

Yeah I could be wrong, but I still think it requires api recognition in said app.  Not sure why it would be slower though.

[Noir Angel]

I'm going to just hope AMD haven't finished optimising their driver yet. I'm going to do a playthrough of Bioshock Infinite and see if that handles any better, I played it enough on W 8.1 to know whether it's handling well or not. The core of the OS seems to be handling fine, and font rendering appears to have been improved even in Chrome. The font rendering generally does feel a lot nicer.

[xendrome]

Font rendering in chrome has nothing to do with Windows 10.

[Luc2k]

So it's not true that it's going to yield gains across the board?

 

On a related note, this is what 3D mark told me 

 

Also, I was reading this and it interested me. Would this make the GPU on my processor worth turning on again, or would I have to wait for Intel and AMD to enable cross GPU support first?

Not yet, unless you have a DX12 game lying around. It would probably require the game to support the feature itself as well.

[George P Global Moderator]

Yeah I could be wrong, but I still think it requires api recognition in said app.  Not sure why it would be slower though.

 

It does, apps/games need to use the new DX12 API to tap into the performance boosts, otherwise it's just going to default to DX11 or w/e the game was written to use.   Remember guys, this is a new low level API, while DX11 cards, and even some older ones, with drivers can use most of the DX12 stuff (other than new features that needs new GPUs) it's all about what the app/game uses.   Older DX11 games will need to get DX12 patches to see performance gains.

[Noir Angel]

Font rendering in chrome has nothing to do with Windows 10.

 

Well the version of Chrome didn't change after the upgrade, and the font rendering is different. So, explanation?

[+Audioboxer Subscriber²]

Seems like one of the OEMS leaked the image MS supply them with. It's floating around on the net.

 

X20-08980_SW_DVD5_NTRL_Win_10_64BIT_English_Win_Pro

 

It includes the windows updates that went live as 10240 launched.

[Xilo]

Upgraded from Windows 7 on my main PC. Working great. Haven't had any issues with games or anything.

[FunkyMike]

Upgraded from Windows 7 on my main PC. Working great. Haven't had any issues with games or anything.

How long did the upgrade take you if I may ask?

[Xilo]

How long did the upgrade take you if I may ask?

Upgrading (keep nothing) took a bit over an hour I would say.

[Noir Angel]

Oddly, although Edge trounces Chrome (even the 64 bit version) in most Javascript benchmarks, it was trounced by Chrome in Peacekeeper (Which oddly recognised Edge as Chrome 42)

[Quillz]

Windows 10 build 10240 RTM upgraded from Windows 7. Thousands on insiders, testers, etc. have been busy prising this Windows 10 mess and nobody noticed this? On the LOGIN SCREEN?

 

One more reason to wait at least 8-10 months before upgrading...a good thing I tested this in a VM instead of my main OS

I have never had this issue happen to me on any Win10 build on either VM or native install.

[Noir Angel]

Doesn't happen on mine either.

[Scott Hellewell]

I just purchased a Dell XPS 13 the other day and successfully upgraded it from from Windows 8.1 to Windows 8.1 Professional and then on to Windows 10 Pro. Things have worked totally smooth and I haven't experienced any problems with it yet.

 

It seems to be pretty speedy.

[HawkMan]

Old games are supposed to get a slight performance boost with DX12 due to cutting a lot of overhead and vastly improved multi core(for both cpu and more importantly GPU) performance. 

 

chances are that that old 3dmark was never meant to take advantage of multi core to start with in which case it's not going to see that performance, on top of that it's DX9 (barely) and not 10 or 11.  DX9 games actually run in a sot of wrapper in DX10/11 again, and had weird performance issues, some got  extra performance in DX10+ some lost performance from the wrapper. 

 

Either way, it fair to say 3dmark05 is an outlier that suffers from several issues causing it to at best not see a DX12 performance boost and at worst a slight decrease. 

[Quillz]

One glitch I do encounter with the Weather app is that it keeps defaulting to Washington, D.C. weather all by itself. I have it set for "detailed" info on the home screen, and after a while, it just goes to that city. I can't figure out how to change the default (back to LA), and even when I thought I did that in the past, it eventually just resets.

[The_Decryptor Veteran]

Oddly, although Edge trounces Chrome (even the 64 bit version) in most Javascript benchmarks, it was trounced by Chrome in Peacekeeper (Which oddly recognised Edge as Chrome 42)

That's because the Edge UA string is an abomination, although what UA isn't at this point.

Old games are supposed to get a slight performance boost with DX12 due to cutting a lot of overhead and vastly improved multi core(for both cpu and more importantly GPU) performance. 

 

chances are that that old 3dmark was never meant to take advantage of multi core to start with in which case it's not going to see that performance, on top of that it's DX9 (barely) and not 10 or 11.  DX9 games actually run in a sot of wrapper in DX10/11 again, and had weird performance issues, some got  extra performance in DX10+ some lost performance from the wrapper. 

 

Either way, it fair to say 3dmark05 is an outlier that suffers from several issues causing it to at best not see a DX12 performance boost and at worst a slight decrease.

I don't really know if it'll help old games much, it'll help older hardware though (Targeting DX11 hardware via DX12 still gives you the driver benefits, but using DX11 to target DX11 hardware won't use the DX12 driver pipeline)

Multi-core support is "ehh" too, previous DX releases handled it internally to the API and it never worked quite right (And a super old release like DX9 simply doesn't have that support, it predates multi-core CPUs). That's one of the nice changes in DX12, leaves multi-threading up to the game/engine devs.

[Noir Angel]

Old games are supposed to get a slight performance boost with DX12 due to cutting a lot of overhead and vastly improved multi core(for both cpu and more importantly GPU) performance. 

 

chances are that that old 3dmark was never meant to take advantage of multi core to start with in which case it's not going to see that performance, on top of that it's DX9 (barely) and not 10 or 11.  DX9 games actually run in a sot of wrapper in DX10/11 again, and had weird performance issues, some got  extra performance in DX10+ some lost performance from the wrapper. 

 

Either way, it fair to say 3dmark05 is an outlier that suffers from several issues causing it to at best not see a DX12 performance boost and at worst a slight decrease. 

 

 

That's because the Edge UA string is an abomination, although what UA isn't at this point.

I don't really know if it'll help old games much, it'll help older hardware though (Targeting DX11 hardware via DX12 still gives you the driver benefits, but using DX11 to target DX11 hardware won't use the DX12 driver pipeline)

Multi-core support is "ehh" too, previous DX releases handled it internally to the API and it never worked quite right (And a super old release like DX9 simply doesn't have that support, it predates multi-core CPUs). That's one of the nice changes in DX12, leaves multi-threading up to the game/engine devs.

 

So they don't fully recognise it's UA string yet. If they're feeding it tests thinking that it's chrome could that also explain the unexpectedly big performance deficit?

 

And the 3d mark 05 test I previously ran was on D3D 11.x on Windows 8.1, I didn't necessarily expect it to gain anything but an 8% performance loss on such an old product just surprised me. Could it just be the app? possibly. I'm about half way into Bioshock Infinite and so far it plays about the same as it did on Win 8.1

[The_Decryptor Veteran]

The tests are the same across browsers, if they aren't then there's not much point to the benchmark.

I'd put any performance differences down to pre-final drivers. The 8 drivers are years old, 10 drivers are a few months old.

[HawkMan]

That's because the Edge UA string is an abomination, although what UA isn't at this point.

I don't really know if it'll help old games much, it'll help older hardware though (Targeting DX11 hardware via DX12 still gives you the driver benefits, but using DX11 to target DX11 hardware won't use the DX12 driver pipeline)

Multi-core support is "ehh" too, previous DX releases handled it internally to the API and it never worked quite right (And a super old release like DX9 simply doesn't have that support, it predates multi-core CPUs). That's one of the nice changes in DX12, leaves multi-threading up to the game/engine devs.

 

DX12 changes the queing of multi threading, that's where most of the performance benefit comes from, besides the bare metal stuff, but that's the stuff that will only give a benefit to DX12 code). preventing stuff like all other cores waiting for the first one, and this on core getting 90% utilization while the others hover at 20-50%. Same thing AMD did with Mantle(though for the CPU only there). the GPU increase here will show everywhere, but like with mantle, high single core performance CPU's like Intel's i5's and up, won't show much benefit in games from this since the single core was more than fast enough to keep up with the needed processing anyway(gamers getting i7's always amuse me) but lower end CPU's will show a market improvement from this. 

[Noir Angel]

The tests are the same across browsers, if they aren't then there's not much point to the benchmark.

I'd put any performance differences down to pre-final drivers. The 8 drivers are years old, 10 drivers are a few months old.

 

I think you mixed me up.

 

The 3D mark 2005 tests were done one on each OS. The peacekeeper results were both done in W10 with the same driver.

[adrynalyne]

So they don't fully recognise it's UA string yet. If they're feeding it tests thinking that it's chrome could that also explain the unexpectedly big performance deficit?

 

 

 

See for yourself:

 

"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/42.0.2311.135 Safari/537.36 Edge/12.10240"?

 

That is read straight from window.navigator.userAgent.

[Gotenks98]

mind posting a brief rig summary?

Operating System: Windows 10 Pro 64-bit (10.0, Build 10240) (10240.th1_st1.150714-1601)

Processor: AMD A10-6800K APU with Radeon HD Graphics   (4 CPUs), ~4.1GHz

Memory: 16384MB RAM

Hard Drive is 256GB crucial SSD cant think of the model right now

Graphics card is HD Radeon 7970