Sign in with
Sign up | Sign in
Your question
Closed

AMD to skip 32nm moving straight to 28nm

Last response: in CPUs
Share
June 24, 2010 1:20:31 PM

Just had a good read through the AMD forums and found a new rumour that AMD is moving straight to 28nm as they were having trouble with 32nm...

Bulldozer

"Zambezi " (28nm, 8-core)

4 AMD Bulldozer modules

Support new ISA: XOP, FMA4, SSSE3, SSE4.1, AVX, AES, CLMUL

Support fusion of CMP and JMP instructions

Advanced Quad-Channel Memory Sub-System

Higher memory level parallelism

Hyper Transport 3.1

AM3r2

10-100 watts

Tentative release date late Q4 2010 to Q1/Q2 2011



"Interlagos" (28nm, 16-core)

8 AMD Bulldozer modules (two dies as MCM)

HyperTransport 3.1

Quad Channel DDR3

Socket G34
a b à CPUs
June 24, 2010 1:44:08 PM

sounds good

Score
0
June 24, 2010 1:50:24 PM

Definitely it sounds awesome, whether it will really happen though?
Score
0
Related resources
Can't find your answer ? Ask !
June 24, 2010 2:04:42 PM

yep that for me was the killer! Now i must wait to upgrade to my new AM3+ motherboard with 8 DDR3 sockets... Take that intel you overpriced bastards!
Score
0
June 25, 2010 7:11:44 AM

Don't hold your breathe, they haven't even tackled a tri channel memory configuration. They are still stuck on dual channel with ddr3. What makes you think they are going to leap frog Intel?
Score
0
June 25, 2010 7:37:08 AM

Hey i'm just repeating what others have said, but yeh quad channel memory would be sweet.
Score
0
a b à CPUs
June 25, 2010 7:52:53 AM

kg4icg said:
Don't hold your breathe, they haven't even tackled a tri channel memory configuration. They are still stuck on dual channel with ddr3. What makes you think they are going to leap frog Intel?


They haven't done Tri channel because there is not enough benefit from tri channel at this juncture. Its not a result of technical difficulties.

That said, 28nm is a half node. So its extremely unlikely that they would do this for a cpu. I think somebody confused an ATI half node 28nm story with an AMD cpu story. The next ITRS defined node is 22nm.

If it was true, it would be an absolutely stunning development.
Score
0
June 25, 2010 9:52:33 AM

Best answer selected by Metallifux.
Score
0
a b à CPUs
June 25, 2010 9:12:29 PM

Quote:
Yeah the quad channel memory has been a rumor for over a year.

I don't think AMD had trouble with 32nm, they just skipped it for obvious reasons, to catch up. They really mastered 45nm to a level Intel could not, this shows in the 6 core cpu's running 125w. Even at 32nm Intels quads are 130w.


They are NOT skipping 32nm. Some rumor on a forum is not evidence.
Score
0
a c 83 à CPUs
a b À AMD
June 25, 2010 9:19:53 PM

kg4icg said:
Don't hold your breathe, they haven't even tackled a tri channel memory configuration. They are still stuck on dual channel with ddr3. What makes you think they are going to leap frog Intel?


Maybe it's the fact the 12 core Opterons have quad channel DDR3 memory.
Score
0
June 26, 2010 11:24:16 AM

FALC0N said:
They are NOT skipping 32nm. Some rumor on a forum is not evidence.



who knows? We will just have to wait for 2 months to find out what AM3+ holds in-store. While i would like to see them push for 28nm i think it is highly unlikely, i think tri or even quad channel ram is highly likely, otherwise they would just stick with AM3.
Score
0
a b à CPUs
June 26, 2010 9:24:12 PM

metallifux said:
who knows? We will just have to wait for 2 months to find out what AM3+ holds in-store. While i would like to see them push for 28nm i think it is highly unlikely, i think tri or even quad channel ram is highly likely, otherwise they would just stick with AM3.


28nm is a half node. The likelyhood of AMD or intel using a half node for a cpu is extremely remote. Its not out of the question, but remote. And when some "forum" post is your source, I think we can safely dismiss it until something better comes along.
Score
0
June 26, 2010 9:55:03 PM

I think 28nm is going to be used for ATI which is currently at 40nm
Score
0
June 27, 2010 1:51:53 AM

what about llano?
Score
0
June 27, 2010 9:38:53 AM

FALC0N said:
when some "forum" post is your source, I think we can safely dismiss it until something better comes along.


It was posted by one of the credible regulars on the official AMD forum so i think it is slightly more reliable than just "some forum".
Score
0
a b à CPUs
June 27, 2010 11:27:21 AM

metallifux said:
It was posted by one of the credible regulars on the official AMD forum so i think it is slightly more reliable than just "some forum".


Sorry but nobody and nothing from "the official AMD forum" that I'm pretty sure you are referring to AMDzone is reliable at all. No regular from there is credible.
Score
0
a b à CPUs
June 27, 2010 12:12:33 PM

if this is true, this will be VERY interesting. the whole reason i am, and a lot of people are, voting for amd to catch up, is either customer loyalty (aka fanboism) or that they want to force all prices down due to competition (theres me). By the sounds of this thing, if intel doesnt release something soon, the bulldozer will eat the 980x for breakfast (speculation at the specs). But then, amd will likely be selling this thing for $1500!!! im sure intel will catch up, then it will be a super high class competition, that will likely leave us mere mortals who cant afford a $1500 cpu in the dust. if amd was smart, which i doubt they will do this, they will sell it for maybe $500, or whatever keeps them in the green. if it beats the 980x, you just screwed it over, now NOBODY would buy it, and u shut intel out of that part of the market. so this could be very interesting if only some of this is true. and quad channel memory doesnt seem like too much of a stretch, they have been operated with 4 memory slots for years, now they just have to mash em together :lol:  but maybe intel will go to hexa channel when amd goes to quad channel?
Score
0
June 28, 2010 4:19:48 AM

If AMD do announce AM3+ socket for bulldozer at this conference would they release it before the actual processor release given that it will likely be compatible with current Phenom II CPU's or will we have to wait till next year? As i would like to build a new machine now just wanna have the most future-proof motherboard.
Score
0
June 29, 2010 6:20:54 PM

yeh Zambezi is what i'm talkin about
Score
0
a c 131 à CPUs
a b À AMD
June 30, 2010 4:48:29 AM

I found this on the "discussion" page of wikipedia:
http://en.wikipedia.org/wiki/List_of_future_AMD_micropr...
http://en.wikipedia.org/wiki/Talk:List_of_future_AMD_mi...
I don't see any sources for what I have bolded. Nor can I find any evidence to support it. I have also never heard of it myself:
What I have put in italics is supposition only. I mean to the point where it should not be included in the article:
"Zambezi" (32nm, 8-core)
4 AMD Bulldozer modules
Support new ISA: XOP, FMA4, SSSE3, SSE4.1, AVX, AES, CLMUL
Support fusion of CMP and JMP instructions
Advanced Quad-Channel Memory Sub-System
Higher memory level parallelism

DDR5 <-- Actually, I just removed this because DDR4 is still in development and was announced as such in 2008.
AM3r2
10-100 watts[2][3]
Tentative release date late Q4 2010 to Q1/Q2 2011
Score
0
June 30, 2010 5:06:25 AM

yep i really don't think zambezi is gonna be built on 28nm but i would put money on am3+ being more than duel channel
Score
0
a b à CPUs
June 30, 2010 5:11:46 AM

Quote:
Yeah the quad channel memory has been a rumor for over a year.

I don't think AMD had trouble with 32nm, they just skipped it for obvious reasons, to catch up. They really mastered 45nm to a level Intel could not, this shows in the 6 core cpu's running 125w. Even at 32nm Intels quads are 130w.

Why would Intel need a hex with 45nm architecture while even their 45nm quad at much lower frequency performs better than 45nm AMD hex on average?
http://www.anandtech.com/bench/Product/146?vs=47

In addition, don't forget that 980x was released before 1090T.

This is obviously a thread created by an AMD fanboy and should stay in AMD section instead.
Score
0
June 30, 2010 6:24:54 AM

iqvl said:
Why would Intel need a hex with 45nm architecture while even their 45nm quad at much lower frequency performs better than 45nm AMD hex on average?
http://www.anandtech.com/bench/Product/146?vs=47

In addition, don't forget that 980x was released before 1090T.

This is obviously a thread created by an AMD fanboy and should stay in AMD section instead.



quite the contrary actually, my intel system is way better than my AMD...
Score
0
July 12, 2010 12:24:18 AM

Globalfoundries 32nm Bulk Process Canceled - Semiconductor maker goes directly to 28nm, similar to TSMC. AMD to skip 32nm moving straight to 28nm.

LINK:
http://en.wikipedia.org/wiki/Bulldozer_(processor)
Globalfoundries Scraps 32nm Bulk Fabrication Process.
http://www.xbitlabs.com/news/other/display/201004011446...

Quote:
AMD Bulldozer Technology

1) Advanced Quad-Channel Memory Sub-System
2) Higher Memory Level Parallelism
3) Extensive New Power Management Innovations
4) Multi-Threading Technology
5) Two 128-bit FMAC Floating Point Units
6) Shared L2 cache
7) Shared L3 cache
8) 28nm SOI with HKMG (32nm cancelled)
9) TDP – 10W to 100W
10) 2 Integer Cores each w/ 4 Pipelines
11) Support new ISA: XOP, FMA4, SSSE3, SSE4.1, AVX, AES, CLMUL
12) Etc...

Intel Sandy Bridge Technology

1) Hyper-Threading Technology
2) 8 MB Shared L3 cache
3) 256 bit/cycle Ring Bus Bandwidth
4) Tri-Channel Integrated Memory Controller
5) Integrated Graphics Core
6) Maximum Thermal Design Power (TDP) of 85W
7) 256-bit FPU
8) DDR3-1600
9) Etc...
Score
0
a b à CPUs
July 12, 2010 12:43:14 AM

nt300 said:
Globalfoundries 32nm Bulk Process Canceled - Semiconductor maker goes directly to 28nm, similar to TSMC. AMD to skip 32nm moving straight to 28nm.

LINK:
http://en.wikipedia.org/wiki/Bulldozer_(processor)
Globalfoundries Scraps 32nm Bulk Fabrication Process.
http://www.xbitlabs.com/news/other/display/201004011446...


Yeah, now look at the references it claims. The only source for AMD skipping 32nm for bulldozer is THIS thread. Its a rumor from people who know nothing about fabs.

There is not one source claiming AMD is using 28nm for cpu. They all reference the bulk process.


Score
0
July 12, 2010 5:43:56 PM

Couldn't see anything about 28nm on Wiki, someone must have removed it already, Falcon? But the xbitlabs is interesting but to me sounds like they are talking more about AMD's GPUs
Score
0
a b à CPUs
July 12, 2010 8:26:53 PM

Its not there anymore. It should never have been there to begin with. It actually linked to this very threat as a source! Ridiculous! I am surprised it was up as long as it was.

They are talking about the bulk process in those articles. 28nm is a standard half node. AMD/ATI GPU's would have been at 28 nm anyways. It is very unlikely that AMD would do a cpu on a half node.

It could be done, but it would have to have in the works for years. It isn't something you could just decide now. AMD says 32nm and there is no reason not to believe them.
Score
0
July 16, 2010 6:29:10 PM

iqvl said:
Why would Intel need a hex with 45nm architecture while even their 45nm quad at much lower frequency performs better than 45nm AMD hex on average?
http://www.anandtech.com/bench/Product/146?vs=47

In addition, don't forget that 980x was released before 1090T.

This is obviously a thread created by an AMD fanboy and should stay in AMD section instead.



lmao you obviously couldn't count, the link you provided shows 1090T beating 920 in 27 out of 42 benchmarks. and almost all heavily threaded benchmarks. at my workplace, our 1090T destroy 920s in graphic rendering. which is all that matters to us. so shutup and sit the fk down.
Score
0
a b à CPUs
July 16, 2010 6:42:04 PM

1090T is also cheaper as a platform, and runs cooler/ less power and so on. just barely, but overall its a better cpu.
Score
0
July 16, 2010 8:06:19 PM

Its the instruction sets which are going to greatly help AMD in encoding and etc., in addition to the dual 128 bit FPUS able to work as one 256 bit. This should help IPC considerably in some areas. Also i thought since the lawsuit (1.25 Bill one) AMD was allowed to use several of Intels instruction sets which they would supposedly implement in BD.
Score
0
August 2, 2010 11:41:50 PM

kg4icg said:
Don't hold your breathe, they haven't even tackled a tri channel memory configuration. They are still stuck on dual channel with ddr3. What makes you think they are going to leap frog Intel?



Apparently this person does not know *** because AMD has already successfully integrated four channel DDR3 memory. Just read up about Magny-Cours.
Score
0
a b à CPUs
August 3, 2010 12:46:07 AM

wittermark said:
lmao you obviously couldn't count, the link you provided shows 1090T beating 920 in 27 out of 42 benchmarks. and almost all heavily threaded benchmarks. at my workplace, our 1090T destroy 920s in graphic rendering. which is all that matters to us. so shutup and sit the fk down.

920 beats 1090T in 26 benchmarks and as fast in 2 cases. LOWER THE BETTER IN SOME BENCHMARKS!
Are you blind? Yes, you must be and so you can't OC 920 as well for huge performance boost.

Furthermore, have a close read at the performance difference between the two in REAL WORLD and SYNTHETIC(IMAGINARY) 7-zip benchmark. They will show you
the fact that 1090T is better only in IMAGINARY tasks.

This is what you are going to see when 920 is OCed to merely 3.3GHz. It's gonna blow away the 1090T at 4.2GHz.
http://www.anandtech.com/bench/Product/146?vs=99

In addition, yes, 1090T's lead in GFX rendering is all that matters to you as a consequence of your ignorance in OCing.

Sigh, full of Newcomer accounts registered for swearing and trolling by existing AMD fanboys.
Score
0
August 3, 2010 1:02:53 AM

iqvl said:
920 beats 1090T in 26 benchmarks and as fast in 2 cases. LOWER THE BETTER IN SOME BENCHMARKS!

Sigh, full of Newcomer accounts registered for swearing and trolling by existing AMD fanboys.


Whats with everyone complaining about minor swearing on these forums? Sure its against TOS, but am i the only one who is not 5?
Score
0
a b à CPUs
August 3, 2010 1:08:47 AM

yannifb said:
Whats with everyone complaining about minor swearing on these forums? Sure its against TOS, but am i the only one who is not 5?

Yes, minor swearing doesn't matter when comes to none AMD fanboy. When it comes to AMD fanboys, AMD fanMODS will come in and ban. Sigh!
Score
0
August 3, 2010 2:36:04 AM

Could you guys explain what the difference is between tri and quad channel memory and what exactly it is?

Thanks
Score
0
August 3, 2010 3:54:19 AM

"[...] that AMD is moving straight to 28nm as they were having trouble with 32nm"

if they're having trouble with 32, they going to have more trouble with 28.

Half-node processes are usually only used on the cache. So the logic would still be 32nm, but the cache would be 28nm. So, AMD is having trouble with 32nm cache?

I can't wait for Bulldozer benchmarks. AMD has lots quite a bit of it's K8 spark.
Score
0
a b à CPUs
August 3, 2010 5:11:43 AM

iqvl said:
Why would Intel need a hex with 45nm architecture while even their 45nm quad at much lower frequency performs better than 45nm AMD hex on average?
http://www.anandtech.com/bench/Product/146?vs=47

what he was saying was if intel was to make a high clock quad core i7, it would consume more than 130w
but amd does high clock on hex (thats n entire two core as oposed to HT) and still keeps it under 125w
dam thats a tough one to decide if intels 45nm is better or not ..... :p 
iqvl said:
920 beats 1090T in 26 benchmarks and as fast in 2 cases. LOWER THE BETTER IN SOME BENCHMARKS!
Are you blind? Yes, you must be and so you can't OC 920 as well for huge performance boost.

Furthermore, have a close read at the performance difference between the two in REAL WORLD and SYNTHETIC(IMAGINARY) 7-zip benchmark. They will show you
the fact that 1090T is better only in IMAGINARY tasks.

This is what you are going to see when 920 is OCed to merely 3.3GHz. It's gonna blow away the 1090T at 4.2GHz.
http://www.anandtech.com/bench/Product/146?vs=99

In addition, yes, 1090T's lead in GFX rendering is all that matters to you as a consequence of your ignorance in OCing.

Sigh, full of Newcomer accounts registered for swearing and trolling by existing AMD fanboys.

intel has a better ipc, no1 is denying that, but in all honesty, in real life daily use of a PC, like u mentioned, how much can speed bump can u feel for the extra bucks on CPU+mobo+extensive cooking+psu for OCing that i7 compared to a stock hex core on a average mobo+stock cooker.
im sure ur single thread application will finish working faster, but do we really want to spend extra lot to finish 7zip compresion half a minute eariler? only real advantage in real life is if u sli.
and lets not forget ur reducing lifescan of ur system as u keep it oc 24/7
if m wrong, please by all means correct me :p 
peace!
nightscope said:
Could you guys explain what the difference is between tri and quad channel memory and what exactly it is?

Thanks

it increases memory bandwidth...therefore can do more at same time.....thats the theory anyway.:p 
Score
0
a b à CPUs
August 3, 2010 5:39:31 AM

@sarwar_r87:

Quote:
what he was saying was if intel was to make a high clock quad core i7, it would consume more than 130w
but amd does high clock on hex (thats n entire two core as oposed to HT) and still keeps it under 125w
dam thats a tough one to decide if intels 45nm is better or not ..... :p 

:non:  :non:  :non:  i7 9xx@3.3GHz consumes about the same power as 1090T@3.2GHz.
http://www.anandtech.com/bench/Product/146?vs=99
In addition, i7-860@95W beats 1090T(125W) to death as well :p  :p  :p 

Quote:
how much can speed bump can u feel for the extra bucks on CPU+mobo+extensive cooking+psu for OCing that i7 compared to a stock hex core on a average mobo+stock cooker.

LGA1156 i7 offers you huge performance gain over 1090T for about the same price.
Oh and even a 200MHz lower clocked i5 750(95W) beats 1055T(125W) X6 in 50%+ cases. :p  :p  :p 

Quote:
im sure ur single thread application will finish working faster, but do we really want to spend extra lot to finish 7zip compresion half a minute eariler?

7-zip single threaded? LMAO :non:  :non:  :non: 
That's 300MB compression. Imagine what will happen when you are trying to compress several 3GB+ files. :p  :p  :p 

Quote:
only real advantage in real life is if u sli.

In addition to this,
Since "outdated" i7 is much better than the "future proof" AMD hex, i7 will last much longer without having to be replaced as soon as 1090T and so it actually saves you more money than AMD platform. :lol:  :lol:  :lol: 
Score
0
a b à CPUs
August 3, 2010 5:58:46 AM

nightscope said:
Could you guys explain what the difference is between tri and quad channel memory and what exactly it is?

Thanks

The higher the number of channels, the wider the bandwidth and so the performance in theory assuming that there's no bottleneck on other parts.
For example, dual(21GB/s) vs tri channel(25.6GB/s)
Score
0
a b à CPUs
August 3, 2010 6:33:02 AM

Wupdy do. I am a big fan of AMD but does anyone actually think this will be a huge leap? 28nm over 32 nm... yay... Intel has a new lineup of 32nms coming out less than a year and AMD still has not released anything below 45nm. Intel will still win the crown regardless of dies.

Want to impress me, AMD? Show me some damn benchmarks.
Score
0
a b à CPUs
August 3, 2010 9:54:57 AM

:hello:  dude what r u smoking?? given the ultra fanboy post of ur, i think ur either high, sleeping, drunk or ur IQ is well below sea level, i would say 300 feet :ouch:  :kaola: 

iqvl said:
@sarwar_r87:

Quote:
what he was saying was if intel was to make a high clock quad core i7, it would consume more than 130w
but amd does high clock on hex (thats n entire two core as oposed to HT) and still keeps it under 125w
dam thats a tough one to decide if intels 45nm is better or not ..... :p 

:non:  :non:  :non:  i7 9xx@3.3GHz consumes about the same power as 1090T@3.2GHz.
http://www.anandtech.com/bench/Product/146?vs=99
In addition, i7-860@95W beats 1090T(125W) to death as well :p  :p  :p 
can u read between posts? ask ur younger bro to help if u have one.

all core i7 9xx are rated at 130W. i920 @ 45nm and 2.6G is at 130W. 860 is at 95W @ 2.8G and 45nm. all are 4 core with HT (which takes like 10% more area)
1090T is a full six core on 45nm @ 3.2G, please note that 130W>125W
what mr. zipzoomflyhigh was trying to say that AMD 45nm is better than intel coz they can get more out of it. and it is true.
i don blame intel for goinf 32nm. but AMD though of two option: 1. go 32, and 2. enhance 45nm. we all know going 32 is costly and needs money amd don have, so they did the other one.
intel on the other hand had that money. so its din bother with 45nm.

now u sir, are too thick to even understand what mr. zipzoomflyhigh wrote and went on blabbing about y intel would need hex on 45nm....which is not what was being discussed. question was whose 45nm was better. although mr. zipzoomflyhigh was wrong to think amd's 32nm would be better because of amd's superior 45nm coz they are hardly the same thing; i was merely pointing out that u have ADD and need to pay attention as to what you read! (in a more polite way!)

:non:  :non: 
Quote:
how much can speed bump can u feel for the extra bucks on CPU+mobo+extensive cooking+psu for OCing that i7 compared to a stock hex core on a average mobo+stock cooker.

LGA1156 i7 offers you huge performance gain over 1090T for about the same price.
Oh and even a 200MHz lower clocked i5 750(95W) beats 1055T(125W) X6 in 50%+ cases. :p  :p  :p 

1st of all, good to see that you are comparing a 1000 dollar cpu with a 295$ one......goes on to show that i was correct about my 1st statement, which you might have taken personally.
however,
lets take i860 which is the same price more or less, ok buddy?
http://www.anandtech.com/bench/Product/146?vs=108
amd wins, or levels 10-11 tests out of 32, that's 33%

but hey, read my post, i did say intel has higher IPC. did u not quote that part intentionally, or did u not know what IPC means so left it out? :o  :pfff: 
nonetheless, lets not 4get we have no upgrade path with 1156, cheaper mobo, and bla bla bla, which any amd fan boy will point out. and any intel fan boy like yourself will refute saying, who upgrades only CPU!

have you ever though companies like amd and intel have started ppl that us average joe working for them? :kaola: 

Quote:
im sure ur single thread application will finish working faster, but do we really want to spend extra lot to finish 7zip compresion half a minute eariler?

7-zip single threaded? LMAO :non:  :non:  :non: 
That's 300MB compression. Imagine what will happen when you are trying to compress several 3GB+ files. :p  :p  :p 

srsy are u high or r u accidentally chose Pentium D vs athlon 64 graphs while comparing?
http://www.anandtech.com/bench/Product/146?vs=108
in winRAR, a 300MB files take intel 80sec to compress. amd 89sec, 9sec or intel is 11% better.
not if it was 3GB, a intel will take 13.3minutes to compress and 14.83 minutes for amd using the same processors.
now i look at 1.53 minues or about 90second differnce in favour of intel for a complete platform that will be about 50$ more expensive......i dono about u, but if i can wait 13 minutes, i would rather wait 1.53 more minutes and pay 50 dollar less, you know. and not be stuck with a 1156 platform for life :kaola: 


Quote:
only real advantage in real life is if u sli.

In addition to this,
Since "outdated" i7 is much better than the "future proof" AMD hex, i7 will last much longer without having to be replaced as soon as 1090T and so it actually saves you more money than AMD platform. :lol:  :lol:  :lol: 

:non:  :non:  :non:  :non: 
future platform are not going to be single thread, intel will make sure of it coz they are spending billions of dollar on RnD not to show off only. future will be multi core. and thats when u know what the benchmark for multi thread looked like for 1090T and 860: http://www.anandtech.com/bench/Product/146?vs=108 ... im not gona ask which one will last longer to you coz i don think ur there yet to be able to reply me


i really confused y m wasting my time on this. i am not a amd fanboy. it is true that out if the 16 years i had a PC at home, i had amd for 15years. not because i like amd or anything. but this is what i do when i want to buy new PC. i set a budget. go to store and check the stuff i can get, both intel and amd platform. come home and read the benchmarks, and everytime i see amd performs better or has similar price with something nice which may be not so imp. (like TDP or better IGP, i feel more secure buying a PC with IGP coz GPU seems to burn out a lot as i live in a hot country)

but i guess i just find it amusing how naive you are. brain washed by intel. either that or your just dumb and insecure coz of all the bulldozer crap.
lets do a math and show u y you feel insecure, coz other wise u would whine about it.
http://www.anandtech.com/bench/Product/147?vs=108
1055T@2.8G vs i7 940 @ 2.93
see the cinebenc single thread performance
1055T@2.8=3457
940@2.93G=4171
also 920@2.66G = 3846
considering linear growth, 940@2.8G would be 4000. if you want u can check with 950@3.06G and you will know that the 4000 is more than fair assumption.
but, you will see that 860@2.8G is 4490
now, thats wierd. but nontheless, lets take 1055T vs 860, shall we since 860 has apperantly better results.

anywho, this means, core i7 is about 23% faster clock for clock. evein if you comapare a i7@2.35 and a 3.0GHz based on assumption that the two perform more or less same in terms of performance, intel is 27% better at clock for clock!. new procesor for intel will probably make that 30%, (new processor is reported to have TDP improvements, not much IPC)

now amd has stated loads of improvements with bulldozer. from the looks of it performance will be double as much. however, lets be fair on you, lets say we will see a similar bump when we moved from K8 to K10.5 ok?. now comparing x2 5600+ and 1055T @ 2.8G on single thread for the same test, amd's IPC improved by 33%............ and BOM! there goes ur IPC advantage of intel, they are on even grounds then. only thing that will probable hold amd back is TDP, but we still have no info on it so we can say!

this is just for me having fun. i wont be buying a new PC anytime soon coz i think an quad @ 3G is good enough. hell im not even OCing, coz i don want to kill my CPU lifespan. enough of day dreaming guys, im off to work! :p 

:love:  flame on :love: 

NB: this is for fun, please don take it personally
NB2: the 33% improvement in amd's IPC is purely assumption. i just want to show how much am d will have to be better to level the game :) 
Score
0
a b à CPUs
August 3, 2010 10:09:32 AM

I love how this is just a straight up discussion about AMD skipping 32nm or not. And we have an Intel Flamer come in and start defending Intel like it's his final fight. Calm down its just a discussion we don't need to make every thread you post on an argument, for gods sake, almost every thread i've seen you post on you turn it into an argument.

Topic-wise it'll be interesting to see the 32nm or 28nm architecture coming from AMD as I really just would like to see how it competes with Intel's seemingly more expensive Sandy Bridge's (1155). Gawd i give a hand to Intel for managing with the i5 750/760 but man the 1155 limiting of OC sure gives AMD an advantage in sales and for things to come from its CPUs performance.
Score
0
a b à CPUs
August 3, 2010 10:37:52 AM

sarwar_r87 said:
:hello:  dude what r u smoking?? given the ultra fanboy post of ur, i think ur either high, sleeping, drunk or ur IQ is well below sea level, i would say 300 feet :ouch:  :kaola: 



i really confused y m wasting my time on this. i am not a amd fanboy. it is true that out if the 16 years i had a PC at home, i had amd for 15years. not because i like amd or anything. but this is what i do when i want to buy new PC. i set a budget. go to store and check the stuff i can get, both intel and amd platform. come home and read the benchmarks, and everytime i see amd performs better or has similar price with something nice which may be not so imp. (like TDP or better IGP, i feel more secure buying a PC with IGP coz GPU seems to burn out a lot as i live in a hot country)

but i guess i just find it amusing how naive you are. brain washed by intel. either that or your just dumb and insecure coz of all the bulldozer crap.
lets do a math and show u y you feel insecure, coz other wise u would whine about it.
http://www.anandtech.com/bench/Product/147?vs=108
1055T@2.8G vs i7 940 @ 2.93
see the cinebenc single thread performance
1055T@2.8=3457
940@2.93G=4171
also 920@2.66G = 3846
considering linear growth, 940@2.8G would be 4000. if you want u can check with 950@3.06G and you will know that the 4000 is more than fair assumption.
but, you will see that 860@2.8G is 4490
now, thats wierd. but nontheless, lets take 1055T vs 860, shall we since 860 has apperantly better results.

anywho, this means, core i7 is about 23% faster clock for clock. evein if you comapare a i7@2.35 and a 3.0GHz based on assumption that the two perform more or less same in terms of performance, intel is 27% better at clock for clock!. new procesor for intel will probably make that 30%, (new processor is reported to have TDP improvements, not much IPC)

now amd has stated loads of improvements with bulldozer. from the looks of it performance will be double as much. however, lets be fair on you, lets say we will see a similar bump when we moved from K8 to K10.5 ok?. now comparing x2 5600+ and 1055T @ 2.8G on single thread for the same test, amd's IPC improved by 33%............ and BOM! there goes ur IPC advantage of intel, they are on even grounds then. only thing that will probable hold amd back is TDP, but we still have no info on it so we can say!

this is just for me having fun. i wont be buying a new PC anytime soon coz i think an quad @ 3G is good enough. hell im not even OCing, coz i don want to kill my CPU lifespan. enough of day dreaming guys, im off to work! :p 

:love:  flame on :love: 

NB: this is for fun, please don take it personally
NB2: the 33% improvement in amd's IPC is purely assumption. i just want to show how much am d will have to be better to level the game :) 

1. Good to know that we none AMD fanboys have much more knowlege in OCing so we don't need to pay for $1000 CPUs but still have $1000 CPUs' performance. :kaola: 

2. A 100-core CPU is useless if it can't beat 4-core ones.
It doesn't matter how many cores you've got. What it does matter is the performance! :D 

3. A 6-core CPU that can't even perform as well as 4-core while still consuming more power is lame in efficiency.
It doesn't matter how many cores you've got. What it does matter is the efficiency! :D 

4. Are you just ignore the fact that "7-zip" is a multi-threaded app and i7 beats 1090T here? More future proof on 1090T?!! Sigh.

5. WOW 33% MAX(instead average) improvement on AMD by adding 50% more cores. Amazing! :lol: 

:pfff:  :pfff:  AMD fanboys just can't think properly and always talk without any knowledge but fantasy.

p.s. You are going to work? Aren't you a professional troll hired by AMD? We don't need you. :hello: 
Score
0
a b à CPUs
August 3, 2010 10:47:26 AM

Many peoples have repeated what I said several times and AMD fanboys just keep ignoring the facts. For example, "We can have Intel EXTREME CPU performance by OCing without having to pay $1000".
Score
0
a b à CPUs
August 3, 2010 11:03:46 AM

If you really wanted EE performance for under $1000 you could just buy a lower-end i7 as well. Of course one should evaluate whether they really need EE performance rather than something that will save them money upfront and (a little) over time with lower power consumption. It's often better to buy something that is 1/4 the price, and upgrade it later. The top end chip of one generation doesn't stay the top performer for long.
Score
0
a b à CPUs
August 3, 2010 12:01:33 PM

iqvl said:
1. Good to know that we none AMD fanboys have much more knowlege in OCing so we don't need to pay for $1000 CPUs but still have $1000 CPUs' performance. :kaola: 

2. A 100-core CPU is useless if it can't beat 4-core ones.
It doesn't matter how many cores you've got. What it does matter is the performance! :D 

3. A 6-core CPU that can't even perform as well as 4-core while still consuming more power is lame in efficiency.
It doesn't matter how many cores you've got. What it does matter is the efficiency! :D 

4. Are you just ignore the fact that "7-zip" is a multi-threaded app and i7 beats 1090T here? More future proof on 1090T?!! Sigh.

5. WOW 33% MAX(instead average) improvement on AMD by adding 50% more cores. Amazing! :lol: 

6. :pfff:  :pfff:  AMD fanboys just can't think properly and always talk without any knowledge but fantasy.

p.s. You are going to work? Aren't you a professional troll hired by AMD? We don't need you. :hello: 



u really are thick! :pfff:  :pfff: 

1. so you are a non amd fanboy, nice! i know you can OC, i never compared a 1000 CPU with a 300 dollar with you. i made no mention of it anywhere. nor did i say amd can OC a lot or say that in the high end AMD is doing good. learn to read for crying out loud :whistle: 
but for the last time, amd is non-existant in the high end. BUT the topic is regarding bulldozer. not deneb/nehelam. i merely stated that at mid end downward, for the price it is not always worth buying intel. if you wana OC then you are not an average jeo, but and enthusiast. u may be paying low, but your PC is high end, somewhere AMD is non existent

2. i do remember intel faniies whiling how P4 with HD was better than athlon 64 back in the days....where were were you....may we missed u back then!

3. maybe its coz we donot read properly, but we were actually taking future bulldozer. and i wasnt comparing performance as much as the fact that AMD has 6 core under 125W where as intel has 6core for 130W but at 65W. and interestingly, before x6, you used to say intel is better coz it can handle 6 threads, now there is no mention of it lol....

4. 7zip is not multi threaded actually. i have the 64bit version installed. just yesterday i unpacked a 8GB starcraft, its single core and even then it barely tax one core. none the less i ignored 7zip coz anandtech din include it in benchmark, they had winRAR!..use some common sense.

5. i was talking about cinebenc single thread performance....what makes you think 2 extra core will help? :pfff:  please read properly and try to grasp the level of discussion instead of trolling. number of core is irrelevant here. i talk about single thread performance because i am talking about IPC. if IPC per core improves, multitaking will improve. then you can compare between the cores

also, its not two exra core. its a quad core with something similar to HT, however, it will have more hardware to it. which is y i gues intel fannies will say its 8 core, but its actually not

6.you may say fantacy, but 33% bump is not unrealistic you know. but then i don expect you to know. and you say im trolling :pt1cable:  :pt1cable: 

cheers :kaola: 
Score
0
a b à CPUs
August 3, 2010 12:27:09 PM

sarwar_r87 said:
u really are thick! :pfff:  :pfff: 

1. so you are a non amd fanboy, nice! i know you can OC, i never compared a 1000 CPU with a 300 dollar with you. i made no mention of it anywhere. nor did i say amd can OC a lot or say that in the high end AMD is doing good. learn to read for crying out loud :whistle: 
but for the last time, amd is non-existant in the high end. BUT the topic is regarding bulldozer. not deneb/nehelam. i merely stated that at mid end downward, for the price it is not always worth buying intel. if you wana OC then you are not an average jeo, but and enthusiast. u may be paying low, but your PC is high end, somewhere AMD is non existent

2. i do remember intel faniies whiling how P4 with HD was better than athlon 64 back in the days....where were were you....may we missed u back then!

3. maybe its coz we donot read properly, but we were actually taking future bulldozer. and i wasnt comparing performance as much as the fact that AMD has 6 core under 125W where as intel has 6core for 130W but at 65W. and interestingly, before x6, you used to say intel is better coz it can handle 6 threads, now there is no mention of it lol....

4. 7zip is not multi threaded actually. i have the 64bit version installed. just yesterday i unpacked a 8GB starcraft, its single core and even then it barely tax one core. none the less i ignored 7zip coz anandtech din include it in benchmark, they had winRAR!..use some common sense.

5. i was talking about cinebenc single thread performance....what makes you think 2 extra core will help? :pfff:  please read properly and try to grasp the level of discussion instead of trolling. number of core is irrelevant here. i talk about single thread performance because i am talking about IPC. if IPC per core improves, multitaking will improve. then you can compare between the cores

also, its not two exra core. its a quad core with something similar to HT, however, it will have more hardware to it. which is y i gues intel fannies will say its 8 core, but its actually not

6.you may say fantacy, but 33% bump is not unrealistic you know. but then i don expect you to know. and you say im trolling :pt1cable:  :pt1cable: 

cheers :kaola: 


1. You can beat the performance of $1000 i7-975 by OCing $280 i7-920.
(Troll is troll. "IT" is close minded and just won't listen to whatever you say.)

2. Did I say Intel is better than Athlon 64? Day dreaming?
(No matter how good AMD was, it is a lame ass AT THE MOMENT.)

3. Yes, AMD supports lame 6-real_thread that loses to Intel 4-real_thread.

4. 7-Zip is multi threaded app, you noob ass!
Proof? TOM: Benchmark Results: Multi-Threaded Applications


5. What we care most is the FINAL PERFORMANCE AND EFFICIENCY instead of IPC, number of cores and power consumption. In addition, IPC on AMD is poorer than its performance.
(Troll is troll. "IT" is close minded and just won't listen to whatever you say.)

6. Did I say 33% boost is unrealistic? Day dreaming? What I said is that "33% MAX(instead of average) boost after adding 50% more cores is lame"
(Troll is troll. "IT" is close minded and just won't listen to whatever you say.)
Score
0
a b à CPUs
August 3, 2010 12:30:09 PM

This is going to go downhill nice and quick.
Score
0
!