Overclock.net › Forums › Industry News › Hardware News › [TT] Leaked slides gives us details about upcoming Haswell-EP
New Posts  All Forums:Forum Nav:

[TT] Leaked slides gives us details about upcoming Haswell-EP - Page 11

post #101 of 109
Quote:
Originally Posted by TranquilTempest View Post

It's true that some tasks are difficult to split into multiple threads, but the vast majority of software isn't anywhere close to that. Take the main loop of a game engine, RTS for example, I see no reason I/O, rendering, pathfinding, unit creation, collision, AI, physics, and unit removal can't run simultaneously, Most of them may have to complete their tasks every iteration, but you can start them all from the state of the previous loop, and update the master game state in a deterministic manner after starting all the tasks. You're still limited by whichever task takes longest, but you know where to optimize. The best reason not to do this is that the target hardware is fast enough to run it all in one or two cores at acceptable framerates.
RTS can benefit greatly from multi-threading. Supreme Commander 1 in 2007 supported at least 8 threads.

However, FPS do not as much. Most processes are dependent on one single function.... real-time user input. So no matter how threaded the game gets, many process still run off user interaction. Being the case, how much does further threading benefit the end-user? If some process requires 1% of the CPU, is it worth throw it off into separate thread?

i.e. BF3 runs about 30-40 threads.
345
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #102 of 109
Quote:
Originally Posted by DuckieHo View Post

RTS can benefit greatly from multi-threading. Supreme Commander 1 in 2007 supported at least 8 threads.
However, FPS do not as much. Most processes are dependent on one single function.... real-time user input. So no matter how threaded the game gets, many process still run off user interaction. Being the case, how much does further threading benefit the end-user? If some process requires 1% of the CPU, is it worth throw it off into separate thread?
i.e. BF3 runs about 30-40 threads.
345
If it's in the same thread as the longest task, yes. I think starting a new thread for an easy to isolate task should be the default procedure, only keep it in the same thread if it helps performance.
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
post #103 of 109
Quote:
Originally Posted by TranquilTempest View Post

If it's in the same thread as the longest task, yes. I think starting a new thread for an easy to isolate task should be the default procedure, only keep it in the same thread if it helps performance.

Spawning new threads always introduces some synchronization code. If an easy to isolate task does not take up significant resources, why introduce additional risk and testing for marginal gain? Programming is an art.

Debugging and testing probably take up more time than coding. Minimizing those phases are important if they do not impact overall performance.
Edited by DuckieHo - 7/5/12 at 11:12am
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #104 of 109
Quote:
Originally Posted by DuckieHo View Post

Spawning new threads always introduces some synchronization code. If an easy to isolate task does not take up significant resources, why introduce additional risk and testing for marginal gain? Programming is an art.
Debugging and testing probably take up more time than coding. Minimizing those phases are important if they do not impact overall performance.
In a large project, a 1% performance gain can be a quite significant find. Isolating a task also makes it easier to test.
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
1
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 930 @ 3.6 Gigabyte x58a ud3r Gigabyte GV-R6870C-1GD Mushkin Redline 
Hard DriveOptical DriveMonitorKeyboard
2x 1TB Spinpoint samsung dvd burner Samsung p2370 + Mitsubishi Diamond Pro 930SB WASD keyboards v1 semi custom w/ cherry browns 
PowerCaseMouse
Antec CP-850 Antec P183 CM Storm Spawn 
  hide details  
Reply
post #105 of 109
Quote:
Originally Posted by TranquilTempest View Post

In a large project, a 1% performance gain can be a quite significant find. Isolating a task also makes it easier to test.

Depends on the project and the task.

Separate thread requires testing on impact if the thread does not responds, gets orphaned, runaway, etc. Replicating bugs also becomes harder. That is why I disagree with the "a new thread for an easy to isolate task should be the default procedure". Besides, how do you define an easy to isolate task?

Programming is an art. wink.gif
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
Once again...
(13 items)
 
  
CPUMotherboardGraphicsRAM
i7 920 [4.28GHz, HT] Asus P6T + Broadcom NetXtreme II VisionTek HD5850 [900/1200] + Galaxy GT240 2x4GB G.Skill Ripjaw X [1632 MHz] 
Hard DriveOSMonitorKeyboard
Intel X25-M 160GB + 3xRAID0 500GB 7200.12 Window 7 Pro 64 Acer H243H + Samsung 226BW XARMOR-U9BL  
PowerCaseMouseMouse Pad
Antec Truepower New 750W Li Lian PC-V2100 [10x120mm fans] Logitech G9 X-Trac Pro 
  hide details  
Reply
post #106 of 109
Quote:
Originally Posted by Alatar View Post

What exactly are you running that requires more than a 4.2ghz 2500K assuming it really is IPC that you want? Also why isn't the chip clocked higher?
And intel didn't really do anything to hinder IB. It still clocks much higher than SB if you have the cooling.
some of my old programs that is poorly threaded. They only run 2 threads and still max out my CPU.

I am sticking to 4.2GHz because it is a step down to the "extreme OC" 4.5-5.0GHz, I need to keep the heat under control and I am expecting this rig to run at least 5yrs retiring it for other purpose later when I got a newer faster rig later. So clocking at extreme level aint good with a typical Hyper 212+ cooler.

as for the IB the TIM does have hinder the already high thermal density IB. Getting the heat off IB is already harder than SB, paste it with cheap TIM are making it even harder. So IB doesnt really offer better IPC/ higher clock over SB when OCed.

I am just MUCH preferred Intel stick to 4 cores on Haswell mainstream and focus on delivering high IPC + high clocks within the TDP. Adding 2 more cores are not helping as it may require Intel to clock the CPU lower to fit the TDP. Intel can have Haswell-E using 6-8 cores, their 4 cores socket 2011 on Sandy cant really sell anyway so it may be just better off making it have 2 cores advantage for next gen.
post #107 of 109
Quote:
Originally Posted by selectstriker2 View Post

I'm curious as to how much programming you have done? You come in here spouting of that software isn't multithreaded because programmers / developers are lazy. Have you ever written multithreaded code? Handled data synchronization, race conditions, thread safe operations?
I can see what you are saying, but efficiency isn't extremely high on the priority list for many application projects.
I also think that if most games weren't being slave driven by big publishers who push out games ahead of when the actual development process is complete we would see better optimized games. I'm sure many simply rely on compiler optimizations.

I've been doing casual programming since the apple IIe was around, so just a little while.
post #108 of 109
Quote:
Originally Posted by DuckieHo View Post

Yet, can you refute the fact that multi-threading increases complexity exponentially as there are more possible state machines? More state machines requires more debugging even with well planned and developed code.
This is a basic CS concept... How much multi-threading development do you do?

I dunno, I've written complex subroutines in assembly since the 6800 was a popular cpu, the real problem is that CS teachers are idiots. Some of the problem is that the threading api on the popular operating systems pretty much blows. But to get a better api, you trade efficiency.

and I don;t do any active developement. But threading is a fancy way of saying routine/subroutine and everything beyond that is pure crap.we have a primary routine, and if your smart, you spin off a few sub rountines, and if you do this properly, its not that bad.

Most programmers would do some good to learn on old cpu were you had to code in assembler.
post #109 of 109
So we looking at no sooner then 2014 for Haswell-EP?
Would that make Ivy Bridge-EP somewhere Q1 of 2013 if they even release an Ivy-EP?
Edited by Seraphic - 7/13/12 at 5:40am
Intel Build
(13 items)
 
AMD Build
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Q9450 at 3.55Ghz Gigabyte GA-X48-DQ6 Sapphire HD 5770 Vapor-X G.SKILL 8GB 1066 DDR2 
Hard DriveOSPowerCase
WD 640GBs 2-Blue 6-Black Windows 7 Ultimate 64-Bit ABS Tagan BZ800 Antec 900 
CPUMotherboardGraphicsRAM
AMD PII X4 720 BE 3.4Ghz BioStar TA790GX 128M Sapphire HD 5770 Crossfire G.SKILL 8GB 1066 DDR2 
Hard DriveOSPowerCase
Twin WD Caviar Blue 640GB Windows 7 Ultimate 64-Bit Antec EA650 650W Antec 300 
  hide details  
Reply
Intel Build
(13 items)
 
AMD Build
(13 items)
 
 
CPUMotherboardGraphicsRAM
Intel Q9450 at 3.55Ghz Gigabyte GA-X48-DQ6 Sapphire HD 5770 Vapor-X G.SKILL 8GB 1066 DDR2 
Hard DriveOSPowerCase
WD 640GBs 2-Blue 6-Black Windows 7 Ultimate 64-Bit ABS Tagan BZ800 Antec 900 
CPUMotherboardGraphicsRAM
AMD PII X4 720 BE 3.4Ghz BioStar TA790GX 128M Sapphire HD 5770 Crossfire G.SKILL 8GB 1066 DDR2 
Hard DriveOSPowerCase
Twin WD Caviar Blue 640GB Windows 7 Ultimate 64-Bit Antec EA650 650W Antec 300 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Hardware News
Overclock.net › Forums › Industry News › Hardware News › [TT] Leaked slides gives us details about upcoming Haswell-EP