Jump to content

threads vs cores

Go to solution Solved by Jurrunio,
37 minutes ago, Pc6777 said:

does it just depend on the program because the split resources with hyperthreading do better in some tasks and suffer more in others?

Exactly. SMT lets tasks use different portions of the core together while the core hops around which task it does next. For example if there's one task that only cares about floating point performance, then because 1 thread is enough to fill up the entire floating point unit of a core, SMT wont help at all. If anything, the time spent on switching tasks can hurt performance.

Ok, so i'm just curious about something, so how good is hyperthreading in terms of performance gain if your using all the cores? like a qaud core cpu vs a quad core with hyper-threading both running a program that can fully utilize 8 threads? I know the one with hyperthreading will do better, not twice as well obviously beacause physical cores are better than virtual cores but how much like 50 percent? would  a 6 core cpu with no hyperthreading have similar prefromance to a quad core with hyperthreding all else equal if running  a program that can utilize 8 or more threads, or does it just depend on the program because the split resources with hyperthreading do better in some tasks and suffer more in others?

Link to comment
Share on other sites

Link to post
Share on other sites

what about games, if a game is highly thread scalable and a quad core is struggling and only getting 20 fps would a quad core with hyper-threading get it from 20 fps up to 30 assuming no gpu bottlneck or would there be a bigger difference in a gaming senerio?

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, Pc6777 said:

what about games, if a game is highly thread scalable and a quad core is struggling and only getting 20 fps would a quad core with hyper-threading get it from 20 fps up to 30 assuming no gpu bottlneck or would there be a bigger difference in a gaming senerio?

Really depends on the game and cpu, look at reviews.

Link to comment
Share on other sites

Link to post
Share on other sites

im not in the 20 fps senerio i just used that as an example, cuz 20-30 is 33 percent and you said about 30 percent, so it depends on if the game is optimized for virtual threads I guess?

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, Pc6777 said:

im not in the 20 fps senerio i just used that as an example, cuz 20-30 is 33 percent and you said about 30 percent, so it depends on if the game is optimized for virtual threads I guess?

Id say most games re much less than 30%. Probably 10% or less if I had to make a guess, but many games are faster with hyperthreading off

Link to comment
Share on other sites

Link to post
Share on other sites

10 minutes ago, Pc6777 said:

 

Most games nowadays work better SMT/HT 

Turning it off will most likely result in performance hits. Only way to find out if the game works better with or without it  is to try it yourself.

Edited by TofuHaroto

PC: Motherboard: ASUS B550M TUF-Plus, CPU: Ryzen 3 3100, CPU Cooler: Arctic Freezer 34, GPU: GIGABYTE WindForce GTX1650S, RAM: HyperX Fury RGB 2x8GB 3200 CL16, Case, CoolerMaster MB311L ARGB, Boot Drive: 250GB MX500, Game Drive: WD Blue 1TB 7200RPM HDD.

 

Peripherals: GK61 (Optical Gateron Red) with Mistel White/Orange keycaps, Logitech G102 (Purple), BitWit Ensemble Grey Deskpad. 

 

Audio: Logitech G432, Moondrop Starfield, Mic: Razer Siren Mini (White).

 

Phone: Pixel 3a (Purple-ish).

 

Build Log: 

Link to comment
Share on other sites

Link to post
Share on other sites

37 minutes ago, Pc6777 said:

does it just depend on the program because the split resources with hyperthreading do better in some tasks and suffer more in others?

Exactly. SMT lets tasks use different portions of the core together while the core hops around which task it does next. For example if there's one task that only cares about floating point performance, then because 1 thread is enough to fill up the entire floating point unit of a core, SMT wont help at all. If anything, the time spent on switching tasks can hurt performance.

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

22 minutes ago, Jurrunio said:

Exactly. SMT lets tasks use different portions of the core together while the core hops around which task it does next. For example if there's one task that only cares about floating point performance, then because 1 thread is enough to fill up the entire floating point unit of a core, SMT wont help at all. If anything, the time spent on switching tasks can hurt performance.

makes sense, tha'ts why fx cpus did so badly an 8 core fx only had 4 fpus, they were a hardware version of hyper-threading with bad ipc, I was asking beacause I just upgraded from an i5 6500 at 3.2 ghz to an i7 6700k at 4.0 ghz (h110m cant overclock) that I found on ebay for an ok price, in games i used to be cpu bound on at ultra settings it seems Im slightly gpu bound now (at ultra settings 1440p nothing to cry about I can turn it down to high) but it also has a 20 percent higher clock speed than the i5 6500. I had a moment of entitlement and researched how the core i3 6100 did with and without hyperhthreading in games, because a lot more games can fully take advantage of 4 threads than 8 threads right now, and well the results were, very surprising, it seems on some games it makes a small to medium sized difference, but some games that are threaded for a lot of cores and arn't build around only having 2 threads in mind, it can launch the frame rate to around double, yes double the fps of no hyperthreading, so when games come out that are build around new intel chips and ryzen 5's and ryzen 7's with ryzen 3's as the min spec without 4 core 4 thread chips in mind, well This new cpu could take me from unplaybe to very playable fps by that logic, I upgraded beacause the next gen concoles have a bunch of cores and threads and this could get me by the first half of the generation before devs start pushing the ps5 or whatever to its limits cpu wise and by that time I will want a gpu upgrade anyway and just build new, so iut makes sense that it would vary, if a game is made for a console with a lot of cores and has a sloppy pc port it would need a certain amount of threads to run properly. 

Link to comment
Share on other sites

Link to post
Share on other sites

26 minutes ago, TofuHaroto said:

Most games nowadays work better SMT/HT 

Turning it off will most likely result in performance hits. Only way to find out if the game works better with or without it  is to try it yourself.

I did the next best thing, looked at i3 6100 fps with it on and off saves times and gives perspective for whats to come. 

Link to comment
Share on other sites

Link to post
Share on other sites

13 hours ago, Pc6777 said:

makes sense, tha'ts why fx cpus did so badly an 8 core fx only had 4 fpus, they were a hardware version of hyper-threading with bad ipc,

No it's not hyperthreading. It still has 2 sets of other stuff for every two cores.

 

More like an 8 core processor with only 4 cores worth of FP power and severe interconnect bottleneck (familiar?)

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

7 hours ago, Jurrunio said:

No it's not hyperthreading. It still has 2 sets of other stuff for every two cores.

 

More like an 8 core processor with only 4 cores worth of FP power and severe interconnect bottleneck (familiar?)

makes sense, i guess i just assumed too much, so its 8 real cores on an fx 8xxx but only 1 fpu per 2 cores bottlenecking the entire chip, so like a grey area between cores and hypertheading that makes it technically count as an entire core but preform worse in fpu heavy tasks, but shine and preform well and not be bottle-necked in non fpu intensive tasks.

Link to comment
Share on other sites

Link to post
Share on other sites

I feel that cpus with hytpethreading will shine in senerios were a drm or multiple drms are running in the background and hogging precious threads and cpu power thats needed to run the game, just a speculation. Or even just windows bs running in the background even though you have no other programs running. Stuff like that dosn't need a whole lot of cpu power, but it will steal clock cycles from programs and games you are trying to run.

Link to comment
Share on other sites

Link to post
Share on other sites

56 minutes ago, Pc6777 said:

but shine and preform well and not be bottle-necked in non fpu intensive tasks.

sort of, sadly FPU task are common for CPUs so it doesnt matter how good the rest of it is. Also the uncore bottleneck is heavy, so the CPU has problems transfering data internally (just like Ryzen :P)

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×