• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

hyperthreading

Easy Rhino

Linux Advocate
Staff member
Joined
Nov 13, 2006
Messages
15,692 (2.31/day)
Location
Mid-Atlantic
System Name Desktop
Processor i5 13600KF
Motherboard AsRock B760M Steel Legend Wifi
Cooling Noctua NH-U9S
Memory 4x 16 Gb Gskill S5 DDR5 @6000
Video Card(s) Gigabyte Gaming OC 6750 XT 12GB
Storage WD_BLACK 4TB SN850x
Display(s) Gigabye M32U
Case Corsair Carbide 400C
Audio Device(s) On Board
Power Supply EVGA Supernova 650 P2
Mouse MX Master 3s
Keyboard Logitech G915 Wireless Clicky
Software Fedora KDE Spin
first, is it hyper threading, or hyperthreading. second, is there a big difference between HT 1000 and HT 2000 ??? anyone have a link to benchmarks between processors with same speed and cache but diff HT ??? id like to investigate this!!!
 
its neither HT is hyperthreading which is made for intels cpus...amd uses hypertransport. they are totally different technologies
 
its neither HT is hyperthreading which is made for intels cpus...amd uses hypertransport. they are totally different technologies

ok so do you know any benchmarks of cpus with same speed, same cache, but different hyper transport rates on amds ??
 
Hyper threading and hyper transport are just catch phrase's for FSB speed higher is always better because the components of the motherboard can communicate with the CPU at a higher speed to put it simply
 
Hyper threading and hyper transport are just catch phrase's for FSB speed higher is always better because the components of the motherboard can communicate with the CPU at a higher speed to put it simply

yea that is pretty much what i figured. but i see a that there is a major price disparity between opteron 1000HT cpus and FX 2000HT cpus. similiar speeds and cache tho. thats why im looking for a good bench showing how the two perform!!!! is the price worth it!!?!?!?!?!
 
in summary; hypertransport 1000 is HT 2000. simply, some manufacturers give its full speed on both the rising and falling edge of the clock cycle, where some just choose to give the speed on the rising edge of the clock cycle. a simpler way to explain it i spose would be 1000 is 2000.

the reason FX chips are so much is because they are multiplier unlocked, yep, thats right, all those extra $$$ really does just buy you an unlocked multiplier.
 
Hyper threading and hyper transport are just catch phrase's for FSB speed higher is always better because the components of the motherboard can communicate with the CPU at a higher speed to put it simply

um......no......they arent the same, HyperThreding is intels "fake dual core" used on nutburst chips, it adds a 2nd FAKE cpu allowing windows to offload some tasks to that, this was done to try and take advantage of the bandwith that nutburst cores are capable of, but it dosnt work so well with some apps, and Windows 2000 HATES it because there isnt a REAL 2nd core.

HyperTransportTech is AMD's replacment for the FSB, its how the chip communicates with external hardware/devices also how amd cores talk to eachother dirrectly(something intel chips still dont do, they gotta use their already saturated FSB)

as ket said, 1000 IS 2000, just a diffrent way of posting it, really for current chips anything 600 and up is PLENTY, no perf hit even on FX dual core chips, HT3(am2+ and am3) wont be usefull for home users because even quadcore isnt gonna saturate the current ht2 buss, really ht3 is ment for server and high end workstations using 4x4 and 4x8 and 4x16 setups(multi quad core chips)
 
um......no......they arent the same, HyperThreding is intels "fake dual core" used on nutburst chips, it adds a 2nd FAKE cpu allowing windows to offload some tasks to that

Well, I would bet it would be called a "fake" CPU in Disneyland. In adultland it's actually advanced technology to use idle parts of the CPU to execute a second thread. Though since there is no additional hardware it isn't as efficient as having a second core. It is however a very handy feature and I hope it comes back as they're saying. Without doubling transistor count and power usage you can still gain a nice performance boost, systems respond a lot faster with it. Sure there is dual core now, but dual core+HTT still is more responsive. I notice that difference between my current system and my previous Xeons.
It is true though that Windows 2000 isn't optimized for it and actually performs worse in quite some cases with it enabled. Windows XP/2003 do correctly use it.

It has completely no relation to Hypertransport or whatsoever. It does nothing similar.
 
well going on what reviews say of the FX chip, it is a beast of an overclocker. still too expensive tho for the fx-60 :twitch:
 
HyperThreding is IMHO only needed for nutburst because the chips by design are so in efficent, HT isnt in current chips because the way they did it with P4 wont work with core2/p6 cores, remmber eading an artical about it, intels working on something simlar that "may" show up in some core3(or whatever they call it) chips, but from what i was hearing all the stuff they tryed to get it working had ended up with the systems using core/pentium-m being slower, its better and easyer to just use 2 real cores then a 2nd logical unit to make a Fake cpu that ends up floging the cache trying to squeeze out some more perf.

HT was done by intel to combat catch misses and true dual core cpu's as well as to try and make better use of a VERY poor cpu design, only an intel fanboi would say that netburst is good, even intel has admited it was a bad move, hence they went back to the pentium-pro(p6) based core thats FAR more efficent and was used in ppro,p2,p3 and pentium-m chips, these chips dont need HT to respond nicely.

i alwase love it when nutburst users talk about responciveness and performance, specly when they then bring HT into the conversation.......it makes me laugh, because the only reasion HT ever came to be was to try and make poorly designed cores perform better and to give intel something to try and counter the x2's.

and 2k dosnt get along with it because it is a FAKE 2nd cpu/core, 2k3 and xp have been patched to see that its a fake 2nd cpu and use it accordingly, xp effictivly is 2k with alot of extra crap dumped on that only newberts really "need", like system restore.....HAHA if you cant fix ur own fookups mayby you should move to an os that dosnt let you fookup ;)

http://en.wikipedia.org/wiki/Hyper-threading

Future

Older Pentium 4 based MPUs use Hyper-Threading, but the current-generation cores, Merom, Conroe and Woodcrest, do not. Hyper-Threading is a specialized form of simultaneous multithreading (SMT), which has been said to be on Intel plans for the generation after Merom, Conroe and Woodcrest.

More recently Hyper-Threading has been branded as energy inefficient. For example, specialist low power CPU design company ARM has stated SMT can use up to 46% more power than dual CPU designs. Furthermore, they claim SMT increases cache thrashing by 42%, whereas dual core results in a 37% decrease[1]. These considerations are claimed to be the reason Intel has dropped SMT from new cores.

However, some low-power chips do still use multithreading, including the PPE from the Cell processor, the CPUs in the Playstation 3, Sun Microsystem's Niagara and the MIPS 34K.[1][citation needed]


p4/nutburst chips need HT because catchmisses really hurt them, long pipes suck ;)

At NGMA's heart is a 14-stage instruction pipeline - around half the length of 'Prescott' pipeline, the same as the old Pentium Pro, and probably in line with 'Dothan' pipeline length. Prescott's pipeline was extended to around 30 stages to support clock frequencies of 4GHz and beyond. Now that Intel is no longer targeting such high clock speeds - thanks to the heat dissipation problem - out goes the need for such a long pipeline, needed to keep the core efficient at high clock speeds.
http://www.theregister.co.uk/2005/08/23/intel_next_gen_architecture/
the articals old but kinda shows why core dont have HT, unlike p4/nutburst chips the pipes are not massivly long, 30+ stages for prescott, ewww, pentium pro/core/core2 have 14.....very diffrent tech, with far less cache misses AND even if you have a cache miss, it dosnt hurt perf neerly a much as it would with LONG pipes because theres less wait time between cache reads :)
 
it makes me laugh, because the only reasion HT ever came to be was to try and make poorly designed cores perform better and to give intel something to try and counter the x2's.

You crack me up little buddy. Hyperthreading has been designed to make servers more efficient, it first appeared in the Foster MP's in 2001. An improved version appeared in Prestonia, because of its success they also enabled it on later P4's. Some years later AMD introduced the X2, I would say it's rather unfair of Intel to battle the X2 long before AMD even developed it.


If you ever want a serious discussion about these matters you could try to name things right instead of mocking everything by using words like "nutburst" and "fanboi" and telling people how you laugh at them.
 
this is all very interesting.
 
yeah ok, and the p4EE isnt just a xeon thats been made for desktop use to ;)

you only saw alot of p4 with HT adds when x2 came out, i even remmber one that implyed it was just as good as having a real dual core, injoy your nutburst core, thankfully they are headed the way of the dodo and intium1 :)
 
yeah ok, and the p4EE isnt just a xeon thats been made for desktop use to ;)

you only saw alot of p4 with HT adds when x2 came out, i even remmber one that implyed it was just as good as having a real dual core, injoy your nutburst core, thankfully they are headed the way of the dodo and intium1 :)

Sigh, HT has been in the P4 logo since the 3.06 and they've been advertizing it ever since. Way before anyone ever even thought about dual core. Though if you honostly want to believe HT is Intels answer to the X2, enjoy mate.
 
yeah ok, and the p4EE isnt just a xeon thats been made for desktop use to ;)

you only saw alot of p4 with HT adds when x2 came out, i even remmber one that implyed it was just as good as having a real dual core, injoy your nutburst core, thankfully they are headed the way of the dodo and intium1 :)

have you ever used a p4 ht system vs. an athlon64 single-cpu system for serious desktop work? the system is a lot snappier with HT
 
dude i got a p4 3gz with ht,em64t and the works next to me, its far slower then my 2.4gz 754 system dispite the 754 being at t2,2.5-3-3-10@400, and the p4 having 2gb ddr2@ 4-4-4-8 at 667mhz, the things SLOW even with your magic HT enabled under windows 2003 server(i dont use xp,its bloated crap)

and dual core was part of the athlon64 design, people been taking about it since amd started posting info about a64, and yes intel has marketed it as an equivlant to dual cores, tho it dosnt really work that way, it just lets them try and work around some of the flaws in the nutburst design.

and yes the p4 is snappyer in windows with HT enabled, but ITS STILL SLOWER THEN MY OTHER AMD BOXES!!!!!!!!
hell my old 2300mhz athlon xp setup(1gb dc ddr400, nf7-s mobo, cheapo video card) and yes thats in heavy windows use, and thats with this p4 having a better hdd and ram then the others. i still love my p3, and my pentium-m laptop i have is good, but neither make up for the fact that intel made the p4.........nasty innefficent peices of crap.......chips got a MASSIVE cooler just to keep it from overheating(things close to 7in tall!!!!!!!)

i dont really care, if you love your p4's and can accept that they are SLOWER then amd's cheaper offerings thats all good, its your choice, personaly i would go crazy if i had to use that p4 630 all the time, my 3500+@3gz is soooo much faster :D
 
amd cores talk to eachother dirrectly(something intel chips still dont do, they gotta use their already saturated FSB) (...)
Check out this article about how different DC processors including K8 and C2D handle intercore communication.
You are in for a surprise.
xbitlabs said:
I can’t find any indication of direct data transfers from one execution core to another in the Athlon 64 X2 processor.
xbitlabs said:
None of the processors with separate caches tested in this review can perform fast data transfers between the cores. Intel’s Core Duo (Yonah) and Conroe, each with a shared L2 cache, are the only processors that ensure fast processing of the same data block by two cores (...)
AshenSugar said:
and yes the p4 is snappyer in windows with HT enabled, but ITS STILL SLOWER THEN MY OTHER AMD BOXES!!!!!!!!
I thought the whole thread was about HT, not how bad P4 is.

And what are all those exclamation marks for? :confused:
 
um......no......they arent the same, HyperThreding is intels "fake dual core" used on nutburst chips, it adds a 2nd FAKE cpu allowing windows to offload some tasks to that, this was done to try and take advantage of the bandwith that nutburst cores are capable of, but it dosnt work so well with some apps, and Windows 2000 HATES it because there isnt a REAL 2nd core.

umm "fake" cpu HTT is the only reason i dont destroy the crummy comps at school i am forced to code on i turned HT on a few and they go much slower its actually a hlaf decent tech and the only reason to buy a P4 cause without HT they are CRAP
 
interesting indeed!
 
The X2 communicate across the HT bus. The older Pen Ds communicate across the chipset's FSB. Core2Duo dual cores communicate across the shared cache. So, in terms of core cross communication Core2>X2>pD. With the Core2 Quad, it's basically 2 Dual core Core2s glued to the same die. The individual dual core communicates across the shared cache, but each dual core communicates with each other across the FSB. Kind of a hack job, but it definitely gets the job done.
 
Wile E said:
The X2 communicate across the HT bus.
Slight correction:
It's across the memory bus, not HT. The non-coherent HT link leads to MCP...
;)
 
Slight correction:
It's across the memory bus, not HT. The non-coherent HT link leads to MCP...
;)
Hmm, my mistake. I always thought the memory bus was considered part of the HT bus.
 
The X2 communicate across the HT bus. The older Pen Ds communicate across the chipset's FSB. Core2Duo dual cores communicate across the shared cache. So, in terms of core cross communication Core2>X2>pD. With the Core2 Quad, it's basically 2 Dual core Core2s glued to the same die. The individual dual core communicates across the shared cache, but each dual core communicates with each other across the FSB. Kind of a hack job, but it definitely gets the job done.

acctualy x2's have an internal crossbar and HT link between cores, and i know about xbit, some of their articals are good, others i really get annoied with, i have emailed them about a few reviews they have done of boards where they didnt update to current bios(current for the time of the review) and thus the boards look worse then they are, or in a few cases better then they are because the maker cut out fetures(like vcore options)

x2,s dont need to share cache, they can talk DIRRECTLY using internal HT corssbar, core2 from what i have read uses the FSB and shared l2 cant be used to communicate between cores.
also opteron chips have an crossbar betwen chips,this is one reasion they outperform intel dual/quad/exct chip setups from many articals i have read, i duno about xbit's artical, really i would rather see real world tests insted of these synthetics.

athlon64/opteron chips also can use the system ram as a shared L3 due to low latancy(thanks IMC).

want some proof that the IMC rox, check the perf of 256k,and even 128k l2 sempy64 chips compared to 512 and 1mb l2 cache chips :)
 
acctualy x2's have an internal crossbar and HT link between cores
That's where I went wrong. I interpreted that as the HT Bus. And from what I understood the Core2s do talk across the cache, but I'll have to do further research, just to make sure, as I could also be mistaken there.
 
I read through this whole thing and find it funny that someone is actually claiming that hyperthreading was an answer to dual core! I loved the days of my P4northwood clocking from 2.6 - 3.6ghz and beating the pants off of all of my buddies normal athlons (3000+) ( cause around that time that is what they were meant to compete with.... and we all know that p4's vs the 64s wasnt a fair fight :P ). And even now with my own x2 4200+ system oc'd to 5000+ ish speeds, that old northwood still is very snappy! I sold the entire system to a friend who is still using it with a 7800gt agp w/ 1gb of ram and the thing runs great. In fact I wish I hadn't sold that northwood setup, not just cause it was my first build :P, but would have been a very good second pc to have around. Of course your AMD boxes will perform better, at least in gaming, cause thats really where HT has no effect, but for everyday usage and multitasking, it worked, IMO, better than the athlons of the day. Not to mention, cooler. Then came prescott....:banghead:
 
Back
Top