Topic Title: Windows 7 vs Windows 8 RTM: Gaming Performance
Topic Summary:
Created On: 10/18/2012 10:27 AM
Status: Post and Reply
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 10/18/2012 10:27 AM
User is offline View Users Profile Print this message

Author Icon
stumped
Wizard

Posts: 7029
Joined: 11/13/2009

http://benchmark3d.com/windows-7-vs-windows-8-rtm-gaming-performance



-------------------------

Intel I7 960 @ 3.87ghz *Intel DX58SO *HIS HD6970 2gb *Corsair TX650M *2x4gb Corsair XMS3 *WD Black 1TB *Windows 7 Home Premium 64bit ***Asus N71Jq laptop *Intel I7 720QM Processor *Mobility Radeon HD5730 1gb *8gb Ram *Windows 7 64bit ** Toshiba P75-A7200 * Intel I7-4700MQ * Windows 8.1 64bit


 * A clear conscience is usually a sign of bad memory *

 10/18/2012 03:55 PM
User is offline View Users Profile Print this message

Author Icon
black_zion
80 Column Mind

Posts: 12189
Joined: 04/17/2008

Did anyone really expect it to be different? Not like the difference between XP and Vista/7, Win8 is basically Windows 7 with a new UI.

-------------------------
ASUS Sabertooth 990FX/Gen3 R2, FX-8350 w/ Corsair H60, 8 GiB G.SKILL RipjawsX DDR3-2133, XFX HD 7970 Ghz, 512GB Vertex 4, 256GB Vector, 240GB Agility 3, Creative X-Fi Titanium w/ Creative Gigaworks S750, SeaSonic X750, HP ZR2440w, Win 7 Ultimate x64
 10/18/2012 04:08 PM
User is offline View Users Profile Print this message

Author Icon
stumped
Wizard

Posts: 7029
Joined: 11/13/2009

...except now you can refer all the Win8 early adopter whinners to this link. Or, read it to them. It won't make any difference to most..it will still be ATI's fault.



-------------------------

Intel I7 960 @ 3.87ghz *Intel DX58SO *HIS HD6970 2gb *Corsair TX650M *2x4gb Corsair XMS3 *WD Black 1TB *Windows 7 Home Premium 64bit ***Asus N71Jq laptop *Intel I7 720QM Processor *Mobility Radeon HD5730 1gb *8gb Ram *Windows 7 64bit ** Toshiba P75-A7200 * Intel I7-4700MQ * Windows 8.1 64bit


 * A clear conscience is usually a sign of bad memory *

 10/18/2012 04:10 PM
User is offline View Users Profile Print this message

Author Icon
black_zion
80 Column Mind

Posts: 12189
Joined: 04/17/2008

Naturally, it's never the game developer's responsibility to fix their own issues

-------------------------
ASUS Sabertooth 990FX/Gen3 R2, FX-8350 w/ Corsair H60, 8 GiB G.SKILL RipjawsX DDR3-2133, XFX HD 7970 Ghz, 512GB Vertex 4, 256GB Vector, 240GB Agility 3, Creative X-Fi Titanium w/ Creative Gigaworks S750, SeaSonic X750, HP ZR2440w, Win 7 Ultimate x64
 10/18/2012 04:15 PM
User is offline View Users Profile Print this message

Author Icon
stumped
Wizard

Posts: 7029
Joined: 11/13/2009

Originally posted by: black_zion Naturally, it's never the game developer's responsibility to fix their own issues

..or Microsoft.



-------------------------

Intel I7 960 @ 3.87ghz *Intel DX58SO *HIS HD6970 2gb *Corsair TX650M *2x4gb Corsair XMS3 *WD Black 1TB *Windows 7 Home Premium 64bit ***Asus N71Jq laptop *Intel I7 720QM Processor *Mobility Radeon HD5730 1gb *8gb Ram *Windows 7 64bit ** Toshiba P75-A7200 * Intel I7-4700MQ * Windows 8.1 64bit


 * A clear conscience is usually a sign of bad memory *

 10/18/2012 05:44 PM
User is offline View Users Profile Print this message

Author Icon
black_zion
80 Column Mind

Posts: 12189
Joined: 04/17/2008

Most often though it is the game developer. Remember Crysis? STALKER (the first one)? And let's not even get into the developers who do direct console ports.

-------------------------
ASUS Sabertooth 990FX/Gen3 R2, FX-8350 w/ Corsair H60, 8 GiB G.SKILL RipjawsX DDR3-2133, XFX HD 7970 Ghz, 512GB Vertex 4, 256GB Vector, 240GB Agility 3, Creative X-Fi Titanium w/ Creative Gigaworks S750, SeaSonic X750, HP ZR2440w, Win 7 Ultimate x64
 10/18/2012 07:36 PM
User is offline View Users Profile Print this message

Author Icon
stumped
Wizard

Posts: 7029
Joined: 11/13/2009

Originally posted by: black_zion Most often though it is the game developer. Remember Crysis? STALKER (the first one)? And let's not even get into the developers who do direct console ports.

Yup....also the classic fubar...RAGE.



-------------------------

Intel I7 960 @ 3.87ghz *Intel DX58SO *HIS HD6970 2gb *Corsair TX650M *2x4gb Corsair XMS3 *WD Black 1TB *Windows 7 Home Premium 64bit ***Asus N71Jq laptop *Intel I7 720QM Processor *Mobility Radeon HD5730 1gb *8gb Ram *Windows 7 64bit ** Toshiba P75-A7200 * Intel I7-4700MQ * Windows 8.1 64bit


 * A clear conscience is usually a sign of bad memory *

 10/18/2012 09:18 PM
User is offline View Users Profile Print this message

Author Icon
k-9
Cookie Monster

Posts: 484
Joined: 05/02/2010

guess it (windows 8) won't run on an fx processor any better than on 7. hopefully we will see some benches on that soon. i'm sure i'm not the only one interested in that



-------------------------

 motherboard gigabyte: ga-ma790fxt-ud5p bios F8h , amd 965 cpu at 4.0 , sileneX efz-120ha4 cooler, 8 gig patriot pgs34g1600llka memory, xfx 750w power supply, xfx 6970 graphics card, 3-22"1080p monitors , coolermaster HAF 922case, windows 7-64

 10/20/2012 09:03 PM
User is offline View Users Profile Print this message

Author Icon
MisterEd
Nerfed

Posts: 60
Joined: 10/10/2012

Originally posted by: k-9 guess it (windows 8) won't run on an fx processor any better than on 7. hopefully we will see some benches on that soon. i'm sure i'm not the only one interested in that

 

Since the AMD FX CPUs have two cores in each module I had read that Windows 7 could not schedule threads efficiently. For example a process with two threads need to run each thread in a different module for best speed. Windows 7 will probably put both threads in the same module. Windows 8 was supposed to handle this better and put threads in different modules first if it can.

Has anyone seen any benchmarks to verify if this problem has been fixed in Windows 8? I have Windows 8 already but do not have an FX processor to try this myself.



-------------------------

Gigabyte GA-990FXA-UD3 | AMD FX-8350 | Corsair H60 Cooler | GSkill RipjawsX (2x4GB) | ASUS GeForce GTX 560 | WD Caviar Blue 1TB | Seagate 750GB | ASUS 24X DVD/RW| Windows 7 Ultimate 64-bit
ASUS M4N82 Deluxe | AMD Phenom II X4 960T | Corsair XMS2(2x2GB) | PNY GeForce GTS 250 (1GB) | Seagate 300GB | Maxtor 200GB/250GB | Memorex 20X DVD/RW | Windows 7 Home Premium 32-bit

 10/21/2012 01:05 AM
User is offline View Users Profile Print this message

Author Icon
Thanny
Elite

Posts: 1296
Joined: 07/13/2009

The optimal scheduling of Bulldozer is really not at all simple. 

Windows 7 has a hotfix available (not sure if it's pushed through Windows Update or not) that changes the scheduling for Bulldozer so that each module is given a thread before the second "core" in any are.  That's a start, but still not optimal.

For one thing, some threads, because they share a lot of data in common, would benefit from running on the same module.  For another, when only two modules are active out of four, the inactive ones can be powered down, which allows the active ones to be automatically overclocked.

Windows 8 is supposed to have a more optimal scheduling algorithm for Bulldozer built-in, though there's much to mitigate against that benefit from what I've seen.

 10/22/2012 09:09 AM
User is offline View Users Profile Print this message

Author Icon
Loui5D
Peon

Posts: 14
Joined: 10/03/2012

I have the feeling they'll not make the newer versions/revisions of directx for win7 soon.



-------------------------

i5-2500k @ 4.2GHz


RoG Maximus V Extreme


Asus DC2HD7950 TOP, Sapphire Vapor-x


850W OCZ


16gb vengeance mem


W7 64bit


Benq Xl2411T

 10/22/2012 09:27 AM
User is offline View Users Profile Print this message

Author Icon
black_zion
80 Column Mind

Posts: 12189
Joined: 04/17/2008

The problem with Bulldozer/Piledriver/Steamroller is there are 2 integer cores but a single shared floating point processor per module. I'll say it again, AMD should have hardwired the integer cores together so it only appears to be a single integer to FP matchup and in the silicon itself code a load balancing algorithm for each module...

-------------------------
ASUS Sabertooth 990FX/Gen3 R2, FX-8350 w/ Corsair H60, 8 GiB G.SKILL RipjawsX DDR3-2133, XFX HD 7970 Ghz, 512GB Vertex 4, 256GB Vector, 240GB Agility 3, Creative X-Fi Titanium w/ Creative Gigaworks S750, SeaSonic X750, HP ZR2440w, Win 7 Ultimate x64
 10/22/2012 11:59 AM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 578
Joined: 08/28/2012

That part wasn't such a bad idea since the majority of x86 instructions are still integer instructions, not FP.  There's also a limit to how much parallelism can be extracted from "average" desktop software, so after a certain point making a chip "wider" turns into a game of diminishing returns.

Most of the problems we see with Bulldozer on the desktop seems to be coming from poor single threaded performance.  If that were fixed, or clock speed could be increased to make up for it, then I'm thinkin that would do a lot of good.  Bulldozer's modules and Intel's hyperthreading are both forms of SMT, so I assume(or at least, I hope) there was a good reason not to piggy-back on the support for hyperthreading that was already there.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.

 10/22/2012 03:22 PM
User is offline View Users Profile Print this message

Author Icon
black_zion
80 Column Mind

Posts: 12189
Joined: 04/17/2008

Except that the same architecture is shared with the Opteron processors as well, and that's where you want high floating point performance. But so many issues have plagued AMD over the past few years. The embarrassment of Brisbane being inferior to Windsor, The TLB errata with first gen Phenom, and Bulldozer being in development for so long before being rushed out before it was truly finished...I'm sure AMD will be able to catch up again one day, but if they are to they will need to get that IPC up drastically, especially as the mobile market is expanding so drastically and AMD starting to enter it soon and the demand for fast, efficient processors right now is favoring Intel and ARM compared to the power hungry AMD offerings, as AMD needs the higher clock rate.

-------------------------
ASUS Sabertooth 990FX/Gen3 R2, FX-8350 w/ Corsair H60, 8 GiB G.SKILL RipjawsX DDR3-2133, XFX HD 7970 Ghz, 512GB Vertex 4, 256GB Vector, 240GB Agility 3, Creative X-Fi Titanium w/ Creative Gigaworks S750, SeaSonic X750, HP ZR2440w, Win 7 Ultimate x64
 10/22/2012 06:00 PM
User is offline View Users Profile Print this message

Author Icon
Mime
Forum Moderator

Posts: 578
Joined: 08/28/2012

It's the same problem with either integer or FP, actually.  That's why we moved from single core to multicore chips in the first place.  If the chip can't increase IPC by extracting more parallel instructions from a single thread, then we'll increase it by executing more than one. 

It just seems odd to me that this scheduling thing happened in the first place.  You can bet both AMD and Microsoft knew about it well before launch, so I assume there was a good reason other than simply not wanting to "follow" Intel that AMD didn't just piggy-back on the support for hyperthreading which was already there.



-------------------------

Do not meddle in the affairs of archers, for they are subtle and quick to anger.
Post Count: +8510
Troll Hunter

The opinions expressed above do not represent those of Advanced Micro Devices or any of their affiliates.



Edited: 10/22/2012 at 06:09 PM by Mime
Statistics
84130 users are registered to the AMD Support and Game forum.
There are currently 4 users logged in.

FuseTalk Hosting Executive Plan v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.