Conroe has limitation on 64 Bit Mode

coldpower27

Golden Member
Jul 18, 2004
1,676
0
76

ted

Member
Oct 9, 1999
51
0
0
Originally posted by: coldpower27
Originally posted by: ted
Conroe not perfect, it has limitation on 64bit mode, please read this url and thread
http://babelfish.altavista.com/babelfis...p%2fdocs%2f2006%2f0718%2fkaigai288.htm
http://www.aceshardware.com/forums/read_post.jsp?id=120062277&forumid=1

While a theoretical problem is interesting.

It reamins to be seen if Conroe will lose performance in 64Bit as opposed to 32Bit. When it becomes a actual problem is when I will have concern.


you want real world benchmark 64bit vs 32, here:
http://akiba.ascii24.com/akiba/column/latestparts/2006/07/14/663447-002.html
 

jose

Platinum Member
Oct 11, 1999
2,079
2
81
So it looks like Conroe has the same pathetic 64bit implementation that plagued Intel's prior chips.

http://www.theinquirer.net/default.aspx?article=16879

I'd say 100% of the Server Market is affected. Most real servers require more than 4 gigs of memory.

This is really to bad, since I was looking to upgrade some of my A64 3500+ systems.(RedHat 64bit OS) But now I'll think I'll pass..

There is really no excuse for Intel not to fix their 64bit extensions, they knew it was broken 2 years ago.

Regards,
Jose
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: jose
So it looks like Conroe has the same pathetic 64bit implementation that plagued Intel's prior chips.

http://www.theinquirer.net/default.aspx?article=16879

I'd say 100% of the Server Market is affected. Most real servers require more than 4 gigs of memory.

This is really to bad, since I was looking to upgrade some of my A64 3500+ systems.(RedHat 64bit OS) But now I'll think I'll pass..

There is really no excuse for Intel not to fix their 64bit extensions, they knew it was broken 2 years ago.

Regards,
Jose

True, and with the up and coming 64bit vista, this could become an issue for home users too.
 

HopJokey

Platinum Member
May 6, 2005
2,110
0
0
Core MA (Conroe, Woodcrest, Merom) does pretty alright with all 64-bit apps in Anandtech server benchmarks: http://www.anandtech.com/IT/showdoc.aspx?i=2793&p=3

The articles are interesting reads however. I don't believe these chips suffer from the same 64-bit problem as the Netburst ones like Jose mentions; the only thing I am sure of in terms of 64 bit performance vs 32 bit performance for these chips is the lack of Macro Ops Fusion for certain 64-bit instructions.
 

dandragonrage

Senior member
Jun 6, 2004
385
0
0
But it still (usually) beats AMD. By the time I'm using 64-bit, I'll probably have my next CPU after Conroe anyway.
 

dmens

Platinum Member
Mar 18, 2005
2,275
965
136
Originally posted by: jose
So it looks like Conroe has the same pathetic 64bit implementation that plagued Intel's prior chips.

http://www.theinquirer.net/default.aspx?article=16879
Regards,
Jose

LOL! That's a nocona, not a conroe. And I'll give you a cookie when you can explain to me how it is broken... or pathetic, for that matter.

As for the conroe rumor, it made no sense whatsoever, so it really sounds like another piss poor "fake 64-bit on intel" rumor.
 

dexvx

Diamond Member
Feb 2, 2000
3,899
0
0
I laugh at all people trying to pathetically interpret that article and what it means.

Anyways, real world performance from several sites on several types of benchmarks indicates on average a minor gain in 64bit mode.

---

On a serious note, the only thing in 64bit-long mode is missing macro-op fusion, but even that might be liimted to certain instructions.

The other issues mentioned were non-conventional (oh noes... 16bit operation penalty in 64bit long mode).
 

zsdersw

Lifer
Oct 29, 2003
10,505
2
0
Originally posted by: jose
So it looks like Conroe has the same pathetic 64bit implementation that plagued Intel's prior chips.

http://www.theinquirer.net/default.aspx?article=16879

I'd say 100% of the Server Market is affected. Most real servers require more than 4 gigs of memory.

This is really to bad, since I was looking to upgrade some of my A64 3500+ systems.(RedHat 64bit OS) But now I'll think I'll pass..

There is really no excuse for Intel not to fix their 64bit extensions, they knew it was broken 2 years ago.

Regards,
Jose


IOMMU cannot be supported in hardware without an integrated memory controller.. which won't come until CSI. Labelling Intel's implementation of x86-64 as "pathetic" is quite.
 

Duvie

Elite Member
Feb 5, 2001
16,215
0
71
I tell you I have to remain consistent on this one...When I had an Intel P4 northwood and the AMd fans were raving about 64bit performance I said it was "meaningless" then...i said I wouldn't even have a 64bit OS in the near future and likely would have migrated to another cpu before the need arose...

bottom line...no completed windows 64bit yet and I am about to be on my 2nd cpu past my first A64 chip....

I think Conroe still won the test eventhough it appears their implementation of 64bit apps was not as good....I will likely still be 2 cpus away form using a 64bit OS and needing it....

I think this is pointless for me...
 

BladeVenom

Lifer
Jun 2, 2005
13,365
16
0
I've looked at so many Conroe benchmarks recently that I can't remember which sight did what, but one of the sites did the benchmarks with both 32 bit and 64 bit software. There was a small difference, but not enough to change the outcome. Conroe was still the best at 64 bits.
 

harpoon84

Golden Member
Jul 16, 2006
1,084
0
0
Originally posted by: BladeVenom
I've looked at so many Conroe benchmarks recently that I can't remember which sight did what, but one of the sites did the benchmarks with both 32 bit and 64 bit software. There was a small difference, but not enough to change the outcome. Conroe was still the best at 64 bits.

www.planetx64.com reviewed Conroe in both WinXP and WinXP64, the conclusion is the same, Conroe beats A64 in both.

www.tech-report.com benchmarked Conroe using both 32bit and 64bit versions of UT2004, Conroe gains the same amount as A64, in fact would you believe it, the P4 gains the most performance by going to 64bit UT2004, however it's still the slowest by far of all the chips.

Oh, and I agree with Duvie, when A64 was released everyone was getting excited thinking all apps and games will turn to 64bit overnight.

3 years down the track and we're still waiting for some meaningful 64 bit apps, and apart from UT2004, 64bit games.
 

dexvx

Diamond Member
Feb 2, 2000
3,899
0
0
What do people expect?

Northwoods had HT technology, which was supposed to pave the way for Dual Core. And there are still many apps that arent SMT/SMP-aware. And those that are are only 2-way. It literally takes a whole development cycle (3-5 years) before this stuff matures.
 

dmens

Platinum Member
Mar 18, 2005
2,275
965
136
Firstly, the decode latency glassjaw doesn't exist as described, nice conjecture though, given the limited info. Secondly, macro fusion is not the only type of fusion, and even with that disabled, it is not a huge hit on performance since 64-bit codes use less OOO resources in general, and the whole point of fusion is to make the machine look more resourceful to begin with.

Case closed.
 

Black69ta

Junior Member
Jul 12, 2006
14
0
0
I know that Woodcrest is based on Core design also, but isn't Itanium II already 64bit enhanced for server/enterprise market. so wouldn't it be logical that woodcrest will be 64bit optimized too, and thus it would be dumb of Intel to spend the money and or transistors to make "home" C2D's blaze in a 64bit environment when none of the software is there to exploit it. And when PC went from 32bit from 16bit didn't 16bit take a "hit" on 32bit systems? So I can see why Intel doesn't care. Like someone said above, it won't be needed for several more generational steps
 

RichUK

Lifer
Feb 14, 2005
10,341
678
126
Originally posted by: Black69ta
I know that Woodcrest is based on Core design also, but isn't Itanium II already 64bit enhanced for server/enterprise market. so wouldn't it be logical that woodcrest will be 64bit optimized too, and thus it would be dumb of Intel to spend the money and or transistors to make "home" C2D's blaze in a 64bit environment when none of the software is there to exploit it. And when PC went from 32bit from 16bit didn't 16bit take a "hit" on 32bit systems? So I can see why Intel doesn't care. Like someone said above, it won't be needed for several more generational steps

Itanium processors are based on Intel's IA64 Instruction Set (based on the RISC Architecture), where as Intel's and AMD's mainstream processors are based on the x86 architecture.

Itanium has to software emulate an x86 working processor in order to run x86 coded software. Aka pretty much all software available today.

Windows and other software vendors have specifically created versions of their software to be ran on the Intel Itanium processors.

Basically this processor has no use to the average user, or mid ranged user either, considering the lack of software support.


 

mhahnheuser

Member
Dec 25, 2005
81
0
0
I don't know about not concerning one's self about 64 bit? I'll be pretty much wanting to move on to 64 Vista on launch, which I hope will be in the next 6 to 8 months. I certainly won't platform change (939 X2) until this has been completed and running smoothly. I think this will be the optimum time to choose the next platform. I'm carrying, as most of you are, 25% of the 4.0 Gb RAM and seriously looking a moving up to 2.0 Gb although i'm weighing up if I should leave that level to DDR2, but given that DDR3's not that far away, I don't know yet? Anyway that RAM address ceiling of 32 bit OS's is comming up quite quickly. Given the right system I personally might be through that ceiling by 2010 if not earlier. That by my reckoning is pretty close.
 

Keysplayr

Elite Member
Jan 16, 2003
21,219
54
91
Originally posted by: RichUK
Originally posted by: Black69ta
I know that Woodcrest is based on Core design also, but isn't Itanium II already 64bit enhanced for server/enterprise market. so wouldn't it be logical that woodcrest will be 64bit optimized too, and thus it would be dumb of Intel to spend the money and or transistors to make "home" C2D's blaze in a 64bit environment when none of the software is there to exploit it. And when PC went from 32bit from 16bit didn't 16bit take a "hit" on 32bit systems? So I can see why Intel doesn't care. Like someone said above, it won't be needed for several more generational steps

Itanium processors are based on Intel's IA64 Instruction Set (based on the RISC Architecture), where as Intel's and AMD's mainstream processors are based on the x86 architecture.

Itanium has to software emulate an x86 working processor in order to run x86 coded software. Aka pretty much all software available today.

Windows and other software vendors have specifically created versions of their software to be ran on the Intel Itanium processors.

Basically this processor has no use to the average user, or mid ranged user either, considering the lack of software support.

I think they call it "EPIC", Rich. Love your sig by the way!! :thumbsup: