Intel Conroe Reviews And Info

Printable View

Show 100 post(s) from this thread on one page

05-20-2006, 09:40 AM
timpanogos

Hey, I was thinking one of those old antique automobiles, shame shame nn_step!! :slapass: (lol)
05-20-2006, 02:43 PM
accord99

Quote:

Originally Posted by DilTech

nn_step is right...
Put an 8 to 16 way intel against an 8 to 16 way AMD and see what occurs... The bandwidth is going to choke the intel system to the point of lights out, while the AMD will take little to no hit due to it's MUCH better interconnect method.

There aren't many 8S Opteron system to compare first as currently, neither HP or Sun make one. There are 8S+ Xeon MP systems from IBM and Unisys. And its false that Opteron takes no hit, past 4S cache coherency traffic and the number of hops between CPUs begins to heavily impact Opteron's scalability.
05-20-2006, 02:49 PM
VulgarHandle

any chance fugger has made enough money to build an 8x opty and an 8x xeon and put it to the test? :D
05-20-2006, 02:56 PM
infrared

Having an insane FSB doesn't always help much. I don't see much difference from 1200mhz to 1340mhz, but upwards of 1066mhz would probably be beneficial to the newer more powerful architectures.
05-20-2006, 03:16 PM
perkam

Is it just me...or do I see a ~$200 cpu with 4MB Cache :eek:

I understand though that it only uses FB-Dimms ? :(

Perkam
05-20-2006, 03:19 PM
nn_step

They need to get the price lower...
05-20-2006, 03:22 PM
onewingedangel

Quote:

Originally Posted by nn_step

They need to get the price lower...

good joke
05-20-2006, 03:25 PM
nn_step

Quote:

Originally Posted by onewingedangel

good joke

What you don't like Cheaper products? :rolleyes:
05-20-2006, 03:47 PM
dicecca112

Quote:

Originally Posted by perkam

Is it just me...or do I see a ~$200 cpu with 4MB Cache :eek:

I understand though that it only uses FB-Dimms ? :(

Perkam

yeah what about the 930D? I've got it speced at 201$ shipped from Newegg
05-20-2006, 03:47 PM
onewingedangel

they're already cheaper than the comparative amd product, and by all indications higher performing. so they're already pretty good value comparatively. No real price premiums over the conroe variants.
05-20-2006, 04:40 PM
Cooper

Quote:

Originally Posted by VulgarHandle

any chance fugger has made enough money to build an 8x opty and an 8x xeon and put it to the test? :D

That would be about $50k for total system ;)
05-20-2006, 05:05 PM
VulgarHandle

Quote:

Originally Posted by Cooper

That would be about $50k for total system ;)

i know ;)
05-20-2006, 05:17 PM
Fred_Pohl

Quote:

Originally Posted by nn_step

They forgot to mention the extreme performance hit that it takes when you go to a Quad SMP setup.

Got a link to back that claim up? All I've seen so far is a 2GHz dual Cloverton (8 cores) outperforming an 8 way 2GHz Opteron in Cinebench. Although the Opteron system did scale 1% better from 1 core to 8 than Cloverton, I wouldn't call that an "extreme performance hit".
05-20-2006, 05:19 PM
Fred_Pohl

Quote:

Originally Posted by nn_step

WoodCrest just got owned by the INQ :rofl:

Why? Because an 8 core Opteron system isn't far behind a 4 core Woodcrest system in Linpack FLOPs?
05-20-2006, 05:38 PM
thunderstruck!

Quote:

Originally Posted by DilTech

nn_step is right...

It's about time:up:
05-20-2006, 06:06 PM
DilTech

Quote:

Originally Posted by accord99

There aren't many 8S Opteron system to compare first as currently, neither HP or Sun make one. There are 8S+ Xeon MP systems from IBM and Unisys. And its false that Opteron takes no hit, past 4S cache coherency traffic and the number of hops between CPUs begins to heavily impact Opteron's scalability.

I know it takes somewhat of a hit, but not much. Definitely nowhere close to intels hit...

Scalingwise the bandwidth bottleneck chokes xeons.
05-20-2006, 06:29 PM
accord99

Quote:

Originally Posted by DilTech

I know it takes somewhat of a hit, but not much. Definitely nowhere close to intels hit...

Scalingwise the bandwidth bottleneck chokes xeons.

The 8S+ Xeon MP platforms also increase memory bandwidth as additional 4S nodes are added.
05-21-2006, 02:39 AM
Pinnacle

This Bad boy is gonna release its fury upon us on June 19th
http://www.theinquirer.net/?article=31843
05-21-2006, 03:18 AM
Fred_Pohl

Quote:

Originally Posted by Pinnacle

This Bad boy is gonna release its fury upon us on June 19th
http://www.theinquirer.net/?article=31843

Pity that the Inq writers can't even keep up with their own rumors. This article contradicts at least two other recent Inq articles. While this one says that Woodcrest's TDP is 80W another recent article claims that Woodcrest's TDP has been lowered to 65W. Where this article says that Intel has merely caught up to Opteron, another recent one says that Woodcrest will own Opteron in the 1 and 2 socket market segment.

BTW, since when does a northbridge take 30W? IIRC a dual socket 3G Woodcrest system draws less power than a dual 2.4G Rev. E Opteron system, even with the northbridge factored in. Rev. F might even that up a bit.
05-21-2006, 04:29 AM
mjp1618

Quote:

Originally Posted by DilTech

nn_step is right...

Intel's server market chips are "t3h suck" when it comes to scaling with multiple chips. Now, while this means nothing to us, it means the world to big companies, and this is why Dell is going AMD for servers.

Put an 8 to 16 way intel against an 8 to 16 way AMD and see what occurs... The bandwidth is going to choke the intel system to the point of lights out, while the AMD will take little to no hit due to it's MUCH better interconnect method.

Intel cancelling CSI essentially killed it in the server market.

Intel has NOT cancelled CSI.

http://www.realworldtech.com/page.cf...05-05-2006#361

Quote:

Tukwila will also feature the debut of the Common Systems Interconnect or CSI. CSI is a low latency, point to point, serial interconnect that uses differential signaling. Tukwila will integrate four full width CSI links and two half width links. Full width links operate at 6.4GT/s or 4.8GT/s in each direction, depending on the SKU. In comparison, current Itanium 2 systems have a 667MT/s bus, that is 128 bits wide for a total of 10.6GB/s of bandwidth. Unfortunately the width of the CSI data path is unknown, so bandwidth estimates are difficult. The most likely scenario is that CSI is 8 or 16 bits wide, which would yield 64 and 128GB/s respectively.

Tukwila also has an on-die CSI router, and cache coherency directories. The router will improve latency for all systems, and the directories should ensure near linear system scalability for large (> 4 socket) systems. It is almost certain that the four full width CSI links will be used for a 2D torus topology, while the half width links will connect to I/O subsystems. This architecture is rather similar to the EV7, which was the first high performance MPU to have an on-die memory controller, router, directories and interconnects. It should hardly be surprising that Intel is following in the footsteps of the EV7, considering that many former DEC architects are now at Intel.

BTW, the article also says that the Tukwila will also have an IMC :)
05-21-2006, 04:47 AM
nn_step

Quote:

Originally Posted by mjp1618

Intel has NOT cancelled CSI.

http://www.realworldtech.com/page.cf...05-05-2006#361

BTW, the article also says that the Tukwila will also have an IMC :)

yeah and it is Itanic based meaning, that only about 20 programs run on it :rolleyes:
05-21-2006, 04:49 AM
BitpowerPM

CSI maybe starts already with the Tigerton MP, the Quad-Core MP Version based on the "Core" microarchitecture concept.

With CSI and 4x3,0 GHz it could be a real competitor to K8L. :rolleyes:

mfg Simon
05-21-2006, 04:56 AM
onewingedangel

tigerton is the replacement to whitefield, the xeon that was tagged to have csi. Due to problems implementing csi, whitefield was scrapped, and tigerton was hobbled together in its place. Its just a dual die FSB based Xeon.

Its likely that the nehelam architecture will be the first to integrate CSI on the desktop front, and is likely a second gen 45nm part (after the shrink and cache increses to the core2 design).
05-21-2006, 05:16 AM
Ailleur

Quote:

Originally Posted by nn_step

yeah and it is Itanic based meaning, that only about 20 programs run on it :rolleyes:

Seriously, how retarded can someone be. I guess you saw the word Montecito in the article and went OMH ITANIC IHIAIAJHHAHA! SUCK!.

The itanium is absolutely awesome at what it does, which is, not running 3dmarks, so i guess thats why you think it does nothing. But theres no contest as to whether or not it was a commercial failure.

Also, i guess when someone actually said CSI was not cancelled, and backed it up with you know, a source, you had to come up with some way to make it look bad. I look forward to the day you back up the stupidities you say with a source.
05-21-2006, 05:24 AM
ORCBEAST

Calm down guyz

NN-Step has been here since AUG 05 and has over 7395+ posts so obviously much of it is bull. Dont worrie its normal :P

What im interested in is the IMC "Integrated Memory Controller"

If Intel can be close to Opteron with there old design and by the benchies Ive seen the new cores are faster with using FSB then imagine how fast with a IMC on it...

Show 100 post(s) from this thread on one page