Overclock.net › Forums › Video Games › PC Gaming › Benchmarks and Answers to: How many cores do I need for a gaming computer? 2, 4, 6, 8? What about Hyper-Threading does it matter?
New Posts  All Forums:Forum Nav:

Benchmarks and Answers to: How many cores do I need for a gaming computer? 2, 4, 6, 8? What about Hyper-Threading does it matter? - Page 5

post #41 of 54
If you like, there's a link it my Sig to some Hyperthreading on/off comparisons I did a few years back with an i7. Might shed some extra light on the matter.
My System
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-2600k @ 4.6ghz Gigabyte Z77X-UD5H Zotac GTX 970 w/ Heatkiller 16GB G.Skill Ares (2133) 
Hard DriveHard DriveCoolingOS
Samsung 830 256GB Samsung 840 500GB XSPC Raystorm / RX240 / UT60 / D5 W7 Home Premium 64 
MonitorKeyboardPowerCase
X-Star DP2710 1440p CM Quickfire TK (browns) Seasonic X-660 NZXT Switch 810 
MouseAudioAudio
Zowie ec1 evo Creative Soundblaster Z Sennheiser HD558 
  hide details  
Reply
My System
(17 items)
 
  
CPUMotherboardGraphicsRAM
i7-2600k @ 4.6ghz Gigabyte Z77X-UD5H Zotac GTX 970 w/ Heatkiller 16GB G.Skill Ares (2133) 
Hard DriveHard DriveCoolingOS
Samsung 830 256GB Samsung 840 500GB XSPC Raystorm / RX240 / UT60 / D5 W7 Home Premium 64 
MonitorKeyboardPowerCase
X-Star DP2710 1440p CM Quickfire TK (browns) Seasonic X-660 NZXT Switch 810 
MouseAudioAudio
Zowie ec1 evo Creative Soundblaster Z Sennheiser HD558 
  hide details  
Reply
post #42 of 54
Realized I hadn't disabled core parking for the HT runs, so:

4 Cores Enabled + Hyperthreading + Core Parking Disabled:
Frames: 5313 - Time: 60000ms - Avg: 88.550 - Min: 72 - Max: 121


4 Cores Enabled + Hyperthreading:
Frames: 5227 - Time: 60000ms - Avg: 87.117 - Min: 66 - Max: 127

4 Cores Enabled:
Frames: 5840 - Time: 60000ms - Avg: 97.333 - Min: 77 - Max: 129

3 Cores Enabled + Hyperthreading + Core Parking Disabled:
Frames: 5278 - Time: 60000ms - Avg: 87.967 - Min: 71 - Max: 118


3 Cores Enabled + Hyperthreading:
Frames: 5061 - Time: 60000ms - Avg: 84.350 - Min: 66 - Max: 122

3 Cores Enabled:
Frames: 4455 - Time: 60000ms - Avg: 74.250 - Min: 61 - Max: 91

2 Cores Enabled + Hyperthreading + Core Parking Disabled:
Frames: 4313 - Time: 60000ms - Avg: 71.883 - Min: 58 - Max: 91


2 Cores Enabled + Hyperthreading:
Frames: 4337 - Time: 60000ms - Avg: 72.283 - Min: 58 - Max: 91

2 Cores Enabled:
Frames: 3364 - Time: 60000ms - Avg: 56.067 - Min: 45 - Max: 74
Quote:
Originally Posted by daviejams View Post

Those FC3 benches , what setting you got it at ? Can't be ultra ...

It is ultra, what were you expecting?
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
post #43 of 54
Pathetic results from Hitman Absolution (same deal with manual 60 second runs, did each one twice this time):

4 Cores Enabled:

Frames: 6487 - Time: 60000ms - Avg: 108.117 - Min: 47 - Max: 185
Frames: 6413 - Time: 60000ms - Avg: 106.883 - Min: 48 - Max: 186

3 Cores Enabled:

Frames: 6114 - Time: 60000ms - Avg: 101.900 - Min: 47 - Max: 166
Frames: 5823 - Time: 60000ms - Avg: 97.050 - Min: 46 - Max: 169

2 Cores Enabled:

Frames: 4605 - Time: 60000ms - Avg: 76.750 - Min: 39 - Max: 117
Frames: 4680 - Time: 60000ms - Avg: 78.000 - Min: 39 - Max: 117

Miserably threaded, and hopelessly CPU bound. Just look at this:



45FPS (with all four cores enabled) despite my ensuring that the GPU was out of the equation.
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
post #44 of 54
Quote:
Originally Posted by TwoCables View Post

I was not thinking that multiple people were mocking me; I just said that you were, and it was from when you went, "Are you beginning to see what I'm saying now, TwoCables?"

It just rubbed me the wrong way, so I decided to stop giving my energy to it.

This is a useful test, 2C, in so far as the stated goals goes. It's not necessarily going to provide a 'universal' answer, that applies to every situation, but in terms of assessing the general ? of whether there are games that seem to be able to take advantage of >2 cores (and if so, at what number of cores do the performance improvements 'max out', and does HT work the same as addt'l cores, or no?) ... it's a sound methodology. Esp. since he's using both Intel and AMD for the tests.

Also important is to test with relatively low gfx settings so you won't 'miss the effect' by making yourself GPU bottlenecked instead. I would also look for areas in the game to test where you tend to see <99% GPU usage (if you can find any) as this generally indicates a CPU constraint (but not always).

Keep in mind also that testing at different CPU clocks (and comparing % change in FPS vs % change in clocks) as well can be useful, because those tests can more positively confirm whether you're being CPU-limited in the particular testing scenario you're running. If +10% cpu clock = + 10% FPS, you know your are 100% CPU limited where you're testing at in the game.

I second the Crysis 3 request ...
Quote:
Originally Posted by Oubadah View Post

Pathetic results from Hitman Absolution (same deal with manual 60 second runs, did each one twice this time):

4 Cores Enabled:

Frames: 6487 - Time: 60000ms - Avg: 108.117 - Min: 47 - Max: 185
Frames: 6413 - Time: 60000ms - Avg: 106.883 - Min: 48 - Max: 186

3 Cores Enabled:

Frames: 6114 - Time: 60000ms - Avg: 101.900 - Min: 47 - Max: 166
Frames: 5823 - Time: 60000ms - Avg: 97.050 - Min: 46 - Max: 169

2 Cores Enabled:

Frames: 4605 - Time: 60000ms - Avg: 76.750 - Min: 39 - Max: 117
Frames: 4680 - Time: 60000ms - Avg: 78.000 - Min: 39 - Max: 117

Miserably threaded, and hopelessly CPU bound. Just look at this:



45FPS (with all four cores enabled) despite my ensuring that the GPU was out of the equation.

Actually, those reflect a fairly nice improvements from the additional cores. You're unlikely to see significantly better results in terms of % improvements in many of these tests. Going from 76 to 108 is a 42% increase in FPS, going from 2 cores to 4 cores .. I bet if you do this same calculation for all the games, you'll discover that's fairly good 'return' overall. You shouldn't expect to see 100% in this case. That MIGHT happen sometimes going from 1 core to 2 cores, but not from 2 to 4. Games are not like video encoding software. wink.gif
Edited by brettjv - 3/21/14 at 6:22pm
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
xeon X5675 6-core @ 4.1ghz (1.29v, 20x205 +ht ) rampage iii extreme msi rx470 gaming X (the $159 budget king) 3 x 2gb corsair xms3 pc12800 (9-9-9-24-1T@1600MHz) 
Hard DriveOptical DriveCoolingOS
hynix 250gb ssd (boot), 2tb deskstar (apps),1tb... plextor px-712sa - still the best optical drive... corsair h8o v2 aio W10 home 
MonitorPowerCaseAudio
asus vw266h 25.5" (1920x1200) abs sl (enermax revolution) * single 70A rail 850w silverstone rv-03 XFi Titanium 
  hide details  
Reply
post #45 of 54
Quote:
Originally Posted by brettjv View Post

Actually, those reflect a fairly nice improvements from the additional cores. You're unlikely to see significantly better results in terms of % improvements in many of these tests. Going from 76 to 108 is a 42% increase in FPS, going from 2 cores to 4 cores ..

Yeah but 3 -> 4 cores is a total non-event. I don't know, I just can't help but be disappointed with this kind of behavior from a 2013 game. And the cost of the engines CPU inefficiency is plain to see: 45fps in a tight SP scenario for God's sake. It's not like this is a case of largely theoretical bottlenecking amplified/only made visible by benching at low a res. Even under normal gaming circumstances this is a big dip below the standard 60Hz refresh rate making it a significant 'real-world' problem.
Edited by Oubadah - 3/21/14 at 6:40pm
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
post #46 of 54
Today I decided to install Crysis 3 for benchmarking:

4 Cores Enabled + Hyperthreading (Parking Disabled):
Frames: 4169 - Time: 60000ms - Avg: 69.483 - Min: 47 - Max: 160
Frames: 4198 - Time: 60000ms - Avg: 69.967 - Min: 50 - Max: 170

4 Cores Enabled:
Frames: 3217 - Time: 60000ms - Avg: 53.617 - Min: 37 - Max: 138
Frames: 3290 - Time: 60000ms - Avg: 54.833 - Min: 33 - Max: 164

3 Cores Enabled
Frames: 2451 - Time: 60000ms - Avg: 40.850 - Min: 21 - Max: 142
Frames: 2343 - Time: 60000ms - Avg: 39.050 - Min: 24 - Max: 128

Wow, I was not expecting those results from Hyperthreading. That's actually a significant difference, one that I could feel the second I entered the map before even looking at the counter. This is completely contradictory to my experiences with Hyperthreading when I tested it with Crysis 3 on a 2600k - in that case it was within the margin of error vs four plain cores. I wonder why it's yielding such significant improvements with the 4770k? Is it that the 4770k has just enough raw power to counteract the negative scaling effect of HT when paired with a highly threaded engine, or is Haswell somehow "better at HT"?

Unfortunately, those are the only benches I'll be taking. I had intended to do 2 core, 3 cores+HT etc, but it just became far too frustrating. I'm experiencing some kind of issue where Origin (I'm going to blame Origin now because it's happened in my Origin Crysis 2 as well) shuffles my saves around. If I try to load a save slot, half the time it loads a completely random save from another part of the game, which is obviously a royal pain in the arse. It was bad enough already that you can't manually save, or specify custom names for saves, or see preview screens of the saves, or even use the ESC menu to load saves (have to exit to full menu), all of which were possible in Crysis in 2007. This is why I hate PC gaming in 2014.
Edited by Oubadah - 3/22/14 at 8:13pm
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
post #47 of 54
Quote:
Originally Posted by Oubadah View Post

Yes, task manager definitely lies. If you run a single Prime 95 thread on two cores, it shows both cores at ~50%, and I think you can see it in my HT graph above. It's funny, because I swear in XP it didn't - 1 Prime95 thread would appear as one taskman core at 100%.

The big question is: does anyone know of a reliable CPU usage reader?
Uhm, because games run on engines...

Task manager doesn't lie. Every X ticks a thread runs the windows scheduler, the scheduler then reschedules it to a free core (could be the one the thread was running on or a different one), or puts it in ready if other threads in the ready queue have higher priority. But it is a little bit inefficient since every time the thread is moved to a different core it has to reload the L1 cache.
Quote:
Originally Posted by givmedew View Post

You can check this out you might like it http://www.cpuid.com/softwares/tmonitor.html I used to use it when overclock AMD Zosma CPUs to ensure that the turbo was still functioning but it makes a pretty decent CPU usage graph... Then you might also want to use http://www.aida64.com/...

as to the why is the OP testing games... because that is the only way to test an engine... besides engines may actually perform differently in different games but for the most case yes how well the game works with multiple cores is going to depend HIGHLY on the this.

The reason I am doing this is to have a real understanding of how much CPUs scale... By the end I would honestly like to know if there are ANY games at ALL that can utilize HT even if it is HT with only 2 real cores... but better yet I would love to see a game that scales to 4 cores with HT and I would like to see at least one game that can utilize 6 or 8 cores... obviously any game that scales to 4 cores with HT will utilize more than 4 cores...

As far as I can tell the reverse scaling is caused by mis-managing the threads... for example putting to important high use threads on one physical processor and then having another physical processor doing almost nothing... It doesn't seem to be a big enough deal to disable HT however if NO other programs that you run benefit from HT then you should disable HT if you can gain even 100MHz from the heat saved.

On my laptop with an i5 480m (2C + HT) , setting dota 2 affinity to 3 cores gave me higher fps than affinity at 2 cores. It's definitly not the case that HT never improves gaming performance.
2013
(11 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 2700k GIGABYTE GA-Z77-D3H Sapphire HD 7870 XT Boost G.SKILL 1866 Sniper 2x4 GB 
Hard DriveHard DriveOSMonitor
Samsung 840 Basic Crucial MX200 Windows 10 HP IPS 23-inch 1080p 
PowerCaseMouse
Be quiet! BN144 System Power 7 500W NZXT Tempest 210 Zowie FK1 
  hide details  
Reply
2013
(11 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core i7 2700k GIGABYTE GA-Z77-D3H Sapphire HD 7870 XT Boost G.SKILL 1866 Sniper 2x4 GB 
Hard DriveHard DriveOSMonitor
Samsung 840 Basic Crucial MX200 Windows 10 HP IPS 23-inch 1080p 
PowerCaseMouse
Be quiet! BN144 System Power 7 500W NZXT Tempest 210 Zowie FK1 
  hide details  
Reply
post #48 of 54
Thread Starter 
Quote:
Task manager doesn't lie. Every X ticks a thread runs the windows scheduler, the scheduler then reschedules it to a free core (could be the one the thread was running on or a different one), or puts it in ready if other threads in the ready queue have higher priority. But it is a little bit inefficient since every time the thread is moved to a different core it has to reload the L1 cache.
Quote:
Originally Posted by givmedew View Post

You can check this out you might like it http://www.cpuid.com/softwares/tmonitor.html I used to use it when overclock AMD Zosma CPUs to ensure that the turbo was still functioning but it makes a pretty decent CPU usage graph... Then you might also want to use http://www.aida64.com/...

as to the why is the OP testing games... because that is the only way to test an engine... besides engines may actually perform differently in different games but for the most case yes how well the game works with multiple cores is going to depend HIGHLY on the this.

The reason I am doing this is to have a real understanding of how much CPUs scale... By the end I would honestly like to know if there are ANY games at ALL that can utilize HT even if it is HT with only 2 real cores... but better yet I would love to see a game that scales to 4 cores with HT and I would like to see at least one game that can utilize 6 or 8 cores... obviously any game that scales to 4 cores with HT will utilize more than 4 cores...

As far as I can tell the reverse scaling is caused by mis-managing the threads... for example putting to important high use threads on one physical processor and then having another physical processor doing almost nothing... It doesn't seem to be a big enough deal to disable HT however if NO other programs that you run benefit from HT then you should disable HT if you can gain even 100MHz from the heat saved.


On my laptop with an i5 480m (2C + HT) , setting dota 2 affinity to 3 cores gave me higher fps than affinity at 2 cores. It's definitly not the case that HT never improves gaming performance.

Thanks for that info... that is what I am trying to find out...

So DOTA 2 will be one of the ones I try with 4 cores then 4 cores with HT on.

Thanks

I want to answer quite a few questions with this... and some have already been answered.

It does seem that game can MOST DEFINATELY take advantage of 4 cores to the extent that even having 3 cores will cause a loss in frame rate. With that information it makes me wonder then does having 6 or 8 cores or does using HT when you already have 4 cores help in some games...

Typically HT on an i7 with 4 cores will cause a small lost in frame rate or the most minute gain. I would like to find at least one game where that is NOT the case... would also like to see a game where more than 4 cores equals better frames with the 8 core AMD set up and then I would also like to test 8 cores at stock speed VS 6 or 4 cores with an OC to see if people are better off disabling cores and getting a higher OC over running more cores at a lower clock... I think in most games the answer will be that a higher clock with less cores wins... but I won't know till we test it.


If anyways has access to the 8 core AMD CPUs free to run some tests... I guess I am more interested in results that show gains from having more cores than results that show no differences... that way I can focus on testing those games...

In the end I won't use your guys results BUT your results will be what tells me what games to test in the first place.... SO THEY ARE BEYOND IMPORTANT!!!

Please keep the info coming.

I should have new benchmarks posted on Tuesday. Sorry for the wait but personal life gets in the way smile.gif
Silence
(20 items)
 
Secondary Rig
(9 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5820k  ASRock x99 extreme 4 AMD R9 290X AMD R9 290X 
RAMHard DriveCoolingCooling
ballistix ddr4 dual channel Samsung 850 EVO 500GB DT 5noz CPU Waterblock XSPC Razor Full Cover GPU Block 
CoolingCoolingCoolingCooling
XSPC Razor Full Cover GPU Block 4x120 Black Ice Xtreme GTX480 Swiftech MCP655 Danger Den Monsoon Pump/Reservoir 
CoolingOSMonitorKeyboard
Corsair Link Fan Controller Windows 8.1 Nixeus 2560x1440 Logitech G15 
PowerMouseMouse PadAudio
Kingwin PF-850 Ttesports Level10M Razer Vespula Sennheiser HD590 
CPUMotherboardGraphicsRAM
Intel i5-2500k gigabyte z68xp-ud4 Gigabyte GTX 970 G1 Gaming DDR3 1600 4x2GB 
Hard DriveCoolingOSMonitor
Western Digital Velociraptor 300GB w/ Intel RST... Corsair H100 Windows 8.1 1920x1200 HP 
Power
PC P+C 1000w 
  hide details  
Reply
Silence
(20 items)
 
Secondary Rig
(9 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5820k  ASRock x99 extreme 4 AMD R9 290X AMD R9 290X 
RAMHard DriveCoolingCooling
ballistix ddr4 dual channel Samsung 850 EVO 500GB DT 5noz CPU Waterblock XSPC Razor Full Cover GPU Block 
CoolingCoolingCoolingCooling
XSPC Razor Full Cover GPU Block 4x120 Black Ice Xtreme GTX480 Swiftech MCP655 Danger Den Monsoon Pump/Reservoir 
CoolingOSMonitorKeyboard
Corsair Link Fan Controller Windows 8.1 Nixeus 2560x1440 Logitech G15 
PowerMouseMouse PadAudio
Kingwin PF-850 Ttesports Level10M Razer Vespula Sennheiser HD590 
CPUMotherboardGraphicsRAM
Intel i5-2500k gigabyte z68xp-ud4 Gigabyte GTX 970 G1 Gaming DDR3 1600 4x2GB 
Hard DriveCoolingOSMonitor
Western Digital Velociraptor 300GB w/ Intel RST... Corsair H100 Windows 8.1 1920x1200 HP 
Power
PC P+C 1000w 
  hide details  
Reply
post #49 of 54
Quote:
Originally Posted by Sisaroth View Post

Task manager doesn't lie. Every X ticks a thread runs the windows scheduler, the scheduler then reschedules it to a free core (could be the one the thread was running on or a different one), or puts it in ready if other threads in the ready queue have higher priority. But it is a little bit inefficient since every time the thread is moved to a different core it has to reload the L1 cache.

It "lies" in the context we were using. That is, you can't draw conclusions about single threaded gaming bottlenecks from task manager.
Quote:
Originally Posted by givmedew View Post

Typically HT on an i7 with 4 cores will cause a small lost in frame rate or the most minute gain. I would like to find at least one game where that is NOT the case.

You already found that game in my last post: Crysis 3 (at least, assuming the i7 is a 4770k). +10FPS on the average, and more importantly, the minimum is no small gain when we're talking about Crysis 3.
Edited by Oubadah - 3/23/14 at 9:32pm
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
Skylake I
(10 items)
 
Skylake II
(9 items)
 
 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 256GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 256GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 Super Flower Leadex Gold 750W 
CaseAudio
Corsair Carbide Air 540 (+3x NF-S12A FLX) Asus Xonar Essence STX 
CPUMotherboardRAMHard Drive
Intel Core i7-6700K 4GHz Gigabyte GA-Z170X-Designare F23a Corsair LPX 32GB CMK32GX4M4A2133C13 Samsung SSD 840 PRO 512GB 
Hard DriveCoolingOSPower
Samsung SSD 840 PRO 512GB Noctua NH-U12S (+NF-S12A PWM) Windows 7 Pro x64 SP1/Windows 10 Pro x64 EVGA SuperNOVA G2 750W 
Case
Silverstone Fortress FT02 (+4x NF-A14 PWM) 
  hide details  
Reply
post #50 of 54
Thread Starter 
Quote:
Originally Posted by Oubadah View Post

Quote:
Originally Posted by Sisaroth View Post

Task manager doesn't lie. Every X ticks a thread runs the windows scheduler, the scheduler then reschedules it to a free core (could be the one the thread was running on or a different one), or puts it in ready if other threads in the ready queue have higher priority. But it is a little bit inefficient since every time the thread is moved to a different core it has to reload the L1 cache.

It "lies" in the context we were using. That is, you can't draw conclusions about single threaded gaming bottlenecks from task manager.
Quote:
Originally Posted by givmedew View Post

Typically HT on an i7 with 4 cores will cause a small lost in frame rate or the most minute gain. I would like to find at least one game where that is NOT the case.

You already found that games in my last post: Crysis 3 (at least, assuming the i7 is a 4770k). +10FPS on the average, and more importantly, the minimum is no small gain when we're talking about Crysis 3.

Oh im sorry

I didn't realize that bro...

thanks for that info...

Man I don't know where to go from hear... I mean I was pretty set on using i5 CPUs... now it looks like the quad core i7 and heck maybe even the 6 core i7 will be noticeably faster...

Thanks for that info...

any more is welcome...

I would say since I have never seen a game that lost a large amount of performance from HT that it would be worth it to get an i7 over an i5 and leave the HT on all the time.
Silence
(20 items)
 
Secondary Rig
(9 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5820k  ASRock x99 extreme 4 AMD R9 290X AMD R9 290X 
RAMHard DriveCoolingCooling
ballistix ddr4 dual channel Samsung 850 EVO 500GB DT 5noz CPU Waterblock XSPC Razor Full Cover GPU Block 
CoolingCoolingCoolingCooling
XSPC Razor Full Cover GPU Block 4x120 Black Ice Xtreme GTX480 Swiftech MCP655 Danger Den Monsoon Pump/Reservoir 
CoolingOSMonitorKeyboard
Corsair Link Fan Controller Windows 8.1 Nixeus 2560x1440 Logitech G15 
PowerMouseMouse PadAudio
Kingwin PF-850 Ttesports Level10M Razer Vespula Sennheiser HD590 
CPUMotherboardGraphicsRAM
Intel i5-2500k gigabyte z68xp-ud4 Gigabyte GTX 970 G1 Gaming DDR3 1600 4x2GB 
Hard DriveCoolingOSMonitor
Western Digital Velociraptor 300GB w/ Intel RST... Corsair H100 Windows 8.1 1920x1200 HP 
Power
PC P+C 1000w 
  hide details  
Reply
Silence
(20 items)
 
Secondary Rig
(9 items)
 
 
CPUMotherboardGraphicsGraphics
Intel i7-5820k  ASRock x99 extreme 4 AMD R9 290X AMD R9 290X 
RAMHard DriveCoolingCooling
ballistix ddr4 dual channel Samsung 850 EVO 500GB DT 5noz CPU Waterblock XSPC Razor Full Cover GPU Block 
CoolingCoolingCoolingCooling
XSPC Razor Full Cover GPU Block 4x120 Black Ice Xtreme GTX480 Swiftech MCP655 Danger Den Monsoon Pump/Reservoir 
CoolingOSMonitorKeyboard
Corsair Link Fan Controller Windows 8.1 Nixeus 2560x1440 Logitech G15 
PowerMouseMouse PadAudio
Kingwin PF-850 Ttesports Level10M Razer Vespula Sennheiser HD590 
CPUMotherboardGraphicsRAM
Intel i5-2500k gigabyte z68xp-ud4 Gigabyte GTX 970 G1 Gaming DDR3 1600 4x2GB 
Hard DriveCoolingOSMonitor
Western Digital Velociraptor 300GB w/ Intel RST... Corsair H100 Windows 8.1 1920x1200 HP 
Power
PC P+C 1000w 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: PC Gaming
  • Benchmarks and Answers to: How many cores do I need for a gaming computer? 2, 4, 6, 8? What about Hyper-Threading does it matter?
Overclock.net › Forums › Video Games › PC Gaming › Benchmarks and Answers to: How many cores do I need for a gaming computer? 2, 4, 6, 8? What about Hyper-Threading does it matter?