H-theading H-tranport...?
Last response: in CPUs
i don't exactly remember what the definition is of each <font color=blue>Hyper-Treading</font color=blue> and <font color=red>Hyper-Transpory</font color=red>
but now there is versions 2.0 of these coming out.
could someone define what each are, and how each of the 2.0 versions will differ.
thanks
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW
but now there is versions 2.0 of these coming out.
could someone define what each are, and how each of the 2.0 versions will differ.
thanks
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW
More about : theading tranport
In other word:
Hyper Threading is one CPU performing like two. One 3.2Ghz = 2x 1.6Ghz (even if the screen says 2x3.2Ghz).
Hyper Transport is like a direct link between the CPU and Ram without going through the usual north bridge (i suppose). This eliminates the bottleneck of slower physical FSB (I suppose).
A fine day!
Hyper Threading is one CPU performing like two. One 3.2Ghz = 2x 1.6Ghz (even if the screen says 2x3.2Ghz).
Hyper Transport is like a direct link between the CPU and Ram without going through the usual north bridge (i suppose). This eliminates the bottleneck of slower physical FSB (I suppose).
A fine day!
Related ressources
- PSU Help - Forum
I think Intel currently released something called "Intel Threading Tools 2.0", which might be the source of the confusion.
Besides, Prescott's implementation of Hyperthreading is somewhat better than Northwood's, so this could be considered a second version, but it has never - to my knowledge - been called HT 2.0. And the improvements in Hyperthreading in Prescott are not so memorable anyway. At least right now; maybe applications using the 2 special instructions will get improved performance...
Hypertransport 2.0, however, is very real and should be launched within the next few months or so. Like Makaveli said, it will be clocked at 1, 1.2 and 1.4Ghz; initial models will be introduced only at 1Ghz, however.
In addition, it is not very likely that that will make a huge difference in overall performance. We've had someone around here with an A64 doing tests of HT bus at 400, 600 and 800Mhz, and it didn't make a difference at all... However, this was hardly very conclusive. Therefore, wether or not the change from 800Mhz to 1Ghz will have a true performance impact is not really clear at this moment...
<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
Besides, Prescott's implementation of Hyperthreading is somewhat better than Northwood's, so this could be considered a second version, but it has never - to my knowledge - been called HT 2.0. And the improvements in Hyperthreading in Prescott are not so memorable anyway. At least right now; maybe applications using the 2 special instructions will get improved performance...
Hypertransport 2.0, however, is very real and should be launched within the next few months or so. Like Makaveli said, it will be clocked at 1, 1.2 and 1.4Ghz; initial models will be introduced only at 1Ghz, however.
In addition, it is not very likely that that will make a huge difference in overall performance. We've had someone around here with an A64 doing tests of HT bus at 400, 600 and 800Mhz, and it didn't make a difference at all... However, this was hardly very conclusive. Therefore, wether or not the change from 800Mhz to 1Ghz will have a true performance impact is not really clear at this moment...
<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>
actually its something i found on a link that you posted not too long ago
<A HREF="http://www.c627627.com/Intel/Pentium4/" target="_new">intel</A>
along side
<A HREF="http://www.c627627.com/AMD/AthlonXP/" target="_new">AMD</A>
march 2004 - Hyper-threading II (SSE3)
Q2 - Hyper-Threading III (TNI)
and in regards to the 3200 HT = 2x1600 is that realy what it is. just cutting the processor in half to make 2 smaller one's... basicly?
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW
<A HREF="http://www.c627627.com/Intel/Pentium4/" target="_new">intel</A>
along side
<A HREF="http://www.c627627.com/AMD/AthlonXP/" target="_new">AMD</A>
march 2004 - Hyper-threading II (SSE3)
Q2 - Hyper-Threading III (TNI)
and in regards to the 3200 HT = 2x1600 is that realy what it is. just cutting the processor in half to make 2 smaller one's... basicly?
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW
>march 2004 - Hyper-threading II (SSE3)
>Q2 - Hyper-Threading III (TNI)
Don't read too much in it. Prescott was rumoured to have better/new hyperthreading (HT), and it might, but it doesnt really show in benchmarks. What Tejas and TNI (Tejas New Instructions) will bring is anyone's guess.
>and in regards to the 3200 HT = 2x1600 is that realy what
>it is. just cutting the processor in half to make 2 smaller
>one's... basicly?
No. Hyperthreading allow a single core to execute two threads "simultaneously". Here is a very good introduction into HT and similar technologies :<A HREF="http://www.arstechnica.com/paedia/h/hyperthreading/hype..." target="_new"> Ars Technica Introduction to Multithreading, Superthreading and Hyperthreading </A>
The results in the case of P4, is a cpu that in many circumstances behaves similar to a dual CPU machine. For instance, one virtual core might be processing folding@home, while the second virtual core is still available to process you other applications. Depending on the applications used, performance is roughly inbetween a single cpu machine (sometimes even slightly worse) and a dual cpu machine. Note that a dual cpu machine is NOT 2x as fast, its overall roughly 30% faster than a single cpu machine, but it has a much better responsiveness when multitasking, or running cpu intensive background apps. On single threaded apps, (like most games), running just one cpu intensive application at once, HT offers virtually no benefit. If you'd be encoding DivX while playing at the same time, HT might be a lot more usefull (framerates won't suffer as much, encoding will also likely be quicker than on a non HT machine). In theory, performance could almost double in such scenario's. in reality, its a decent and noticeable increase. No more, but also no less.
= The views stated herein are my personal views, and not necessarily the views of my wife. =
>Q2 - Hyper-Threading III (TNI)
Don't read too much in it. Prescott was rumoured to have better/new hyperthreading (HT), and it might, but it doesnt really show in benchmarks. What Tejas and TNI (Tejas New Instructions) will bring is anyone's guess.
>and in regards to the 3200 HT = 2x1600 is that realy what
>it is. just cutting the processor in half to make 2 smaller
>one's... basicly?
No. Hyperthreading allow a single core to execute two threads "simultaneously". Here is a very good introduction into HT and similar technologies :<A HREF="http://www.arstechnica.com/paedia/h/hyperthreading/hype..." target="_new"> Ars Technica Introduction to Multithreading, Superthreading and Hyperthreading </A>
The results in the case of P4, is a cpu that in many circumstances behaves similar to a dual CPU machine. For instance, one virtual core might be processing folding@home, while the second virtual core is still available to process you other applications. Depending on the applications used, performance is roughly inbetween a single cpu machine (sometimes even slightly worse) and a dual cpu machine. Note that a dual cpu machine is NOT 2x as fast, its overall roughly 30% faster than a single cpu machine, but it has a much better responsiveness when multitasking, or running cpu intensive background apps. On single threaded apps, (like most games), running just one cpu intensive application at once, HT offers virtually no benefit. If you'd be encoding DivX while playing at the same time, HT might be a lot more usefull (framerates won't suffer as much, encoding will also likely be quicker than on a non HT machine). In theory, performance could almost double in such scenario's. in reality, its a decent and noticeable increase. No more, but also no less.
= The views stated herein are my personal views, and not necessarily the views of my wife. =
ok yea thanks that more like what i thought H-threading was.
To sum up
<font color=blue>HyperTheading II</font color=blue> next generation is a negligable improvement so far, useing only a cuple new instructions
<font color=red>Hyper-Transport Bus 2.0</font color=red> is simply a jump in bus speeds from 800Mhz to 1.0-1.4Mhz
aslo HT and HT Bus is how they should be writen...correct?
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW<P ID="edit"><FONT SIZE=-1><EM>Edited by cdpage on 02/11/04 03:59 PM.</EM></FONT></P>
To sum up
<font color=blue>HyperTheading II</font color=blue> next generation is a negligable improvement so far, useing only a cuple new instructions
<font color=red>Hyper-Transport Bus 2.0</font color=red> is simply a jump in bus speeds from 800Mhz to 1.0-1.4Mhz
aslo HT and HT Bus is how they should be writen...correct?
<b><A HREF="http://www.1112.net/lastpage.html" target="_new">Finally...the LAST PAGE</A></b>
ASUS P4S8X-P4 2.4B - 2x512M DDR333 - ATI 9500Pro - WD80G HD(8M) - SAMSUNG SV0844D 8G HD - LG 16X DVD - Yamaha F1 CDRW<P ID="edit"><FONT SIZE=-1><EM>Edited by cdpage on 02/11/04 03:59 PM.</EM></FONT></P>
>HyperTheading II next generation is a negligable
>improvement so far, useing only a cuple new instructions
Instruction count doesnt make or break anything. But so far, HT II has not brought substantial benefits over HT-I.
>Hyper-Transport Bus 2.0 is simply a jump in bus speeds from
>800Mhz to 1.0-1.4Mhz
Its more than that; it allows seamless integration of PCI express as well. For desktops, that will be more important (I guess) than increased speed. There will be no more need for a Northbridge on AMD cpu's. The memory controller is built in, and the PCI Express videocards will be hooked up directly to the HT tunnels towards the cpu as well (if I understood that part correctly). That means lower cost, and possibly, better performance.
>aslo HT and HT Bus is how they should be writen...correct?
I like to use HTT for HyperTransportT and HT for HyperThreading. But that is just me. I'm more worried about an abbreviation for PCI Express, as PCI-X is a completely differnt protocol.. PCI-E I suppose ?
= The views stated herein are my personal views, and not necessarily the views of my wife. =
>improvement so far, useing only a cuple new instructions
Instruction count doesnt make or break anything. But so far, HT II has not brought substantial benefits over HT-I.
>Hyper-Transport Bus 2.0 is simply a jump in bus speeds from
>800Mhz to 1.0-1.4Mhz
Its more than that; it allows seamless integration of PCI express as well. For desktops, that will be more important (I guess) than increased speed. There will be no more need for a Northbridge on AMD cpu's. The memory controller is built in, and the PCI Express videocards will be hooked up directly to the HT tunnels towards the cpu as well (if I understood that part correctly). That means lower cost, and possibly, better performance.
>aslo HT and HT Bus is how they should be writen...correct?
I like to use HTT for HyperTransportT and HT for HyperThreading. But that is just me. I'm more worried about an abbreviation for PCI Express, as PCI-X is a completely differnt protocol.. PCI-E I suppose ?
= The views stated herein are my personal views, and not necessarily the views of my wife. =
Quote:
In other word:Hyper Threading is one CPU performing like two. One 3.2Ghz = 2x 1.6Ghz (even if the screen says 2x3.2Ghz).
now you know that isnt true...
-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
Bandwith requirements for a single cpu desktop hardly seem to require faster implementations. SATA doesnt even come close, not even when using 10 disks in a stripe set. DDR doesnt interface with HTT at all, its hooked up directly to the cpu (memory controller).
No, the main advantage of the higher bandwith of HTT2 will be multiprocessor systems (and off course embedded implementations such as routers, etc,..). Faster cpu interconnect will probably give 4+ way opterons another very nice and cheap performance boost.
= The views stated herein are my personal views, and not necessarily the views of my wife. =
No, the main advantage of the higher bandwith of HTT2 will be multiprocessor systems (and off course embedded implementations such as routers, etc,..). Faster cpu interconnect will probably give 4+ way opterons another very nice and cheap performance boost.
= The views stated herein are my personal views, and not necessarily the views of my wife. =
yea.. i think we got another "ill just make it up because it sounds technically inclined" troll again
wheres SoD when you need him/
-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
wheres SoD when you need him/
-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
!
