Conroe 2.4GHz benchmarked

[996GT2]

I know this was on page 5 of the "Intel's conroe in the hands..." thread, but I posted this just to make sure everyone can easily see it

The 3DMARK score has been fixed; it is not 3008 like in the screenshot. He installed a 6600LE and got 8000+
http://www.xtremesystems.org/forums/showthread.php?t=95021&page=1

EDIT: FIxed link
Fixed 3DMARK Score posted by Absolute0:
http://www.xtremesystems.org/forums/showpost.php?p=1376872&postcount=476

 

[furballi]

Benchmarks are fun, but they don't predict real world performance. My 3100 Sempron does 38 seconds at 2.44GHz. I'm sure Conroe isn't 70% faster with all my applications. Today's bottlenecks are the RAMs, HDDs, and optical drives.

Folks always use benchmarks to measure PC performance because there is very little difference speed wise among today's PCs, especially with the majority of productivity applications.

 

[Cooler]

Originally posted by: 996GT2
I know this was on page 5 of the "Intel's conroe in the hands..." thread, but I posted this just to make sure everyone can easily see it

http://www.xtremesystems.org/forums/showthread.php?t=95021&page=1

EDIT: FIxed link

this the omg of the week!

 

[Shimmishim]

Originally posted by: furballi
Benchmarks are fun, but they don't predict real world performance. My 3100 Sempron does 38 seconds at 2.44GHz. I'm sure Conroe isn't 70% faster with all my applications. Today's bottlenecks are the RAMs, HDDs, and optical drives.

Folks always use benchmarks to measure PC performance because there is very little difference speed wise among today's PCs, especially with the majority of productivity applications.

do you know how amazing 21.5 seconds is at spi 1m?

a 3.8 to 3.9 ghz single core FX or 64 chip does around 20 seconds at spi 1M.

you have a dual core chip at 2.4 ghz doing 21.5 seconds. i could only imagine what sort of spi results you'd get at 3 ghz.

either way this chip is going to be A LOT faster than you 2.44 ghz sempron hands down. maybe not 70%.

 

[RichUK]

Originally posted by: Shimmishim

Originally posted by: furballi
Benchmarks are fun, but they don't predict real world performance. My 3100 Sempron does 38 seconds at 2.44GHz. I'm sure Conroe isn't 70% faster with all my applications. Today's bottlenecks are the RAMs, HDDs, and optical drives.

Folks always use benchmarks to measure PC performance because there is very little difference speed wise among today's PCs, especially with the majority of productivity applications.

do you know how amazing 21.5 seconds is at spi 1m?

a 3.8 to 3.9 ghz single core FX or 64 chip does around 20 seconds at spi 1M.

you have a dual core chip at 2.4 ghz doing 21.5 seconds. i could only imagine what sort of spi results you'd get at 3 ghz.

either way this chip is going to be A LOT faster than you 2.44 ghz sempron hands down. maybe not 70%.

exactly, plus it has an inferior memory bus compared to AMD?s, which makes it that little more amazing. Although as with the Dothans with their 2MB's L2 cache, Super pi probably ran totally from the L2 cache when running just 1m. I would like to see a 32m run.

 

[Furen]

All I have to say is 4MB L2. Going to 32M would require the chip to hit the ram which would affect performance tremendously.

 

[RichUK]

Originally posted by: Furen
All I have to say is 4MB L2. Going to 32M would require the chip to hit the ram which would affect performance tremendously.

Exactly :thumbsup:

 

[Duvie]

It is great score but those morons at XS only live by SuperPi...Fugger is only interested in wolrd records and less actual real world apps...

This great score could have more to do with the new cache and buffers then actual IPC speed...That is well above the 20% conroe was leading in real world apps that often dont count on the cache as much....

If you look at it the P4's and the netburst had done quite well in superpi over the last few years and we know where ther were landing in real world apps....

So in conclusion once those guys stop hand jacking themselves over worthless benches and start running real world apps ...LET ME KNOW

 

[Markfw]

Originally posted by: Duvie
It is great score but those morons at XS only live by SuperPi...Fugger is only interested in wolrd records and less actual real world apps...

This great score could have more to do with the new cache and buffers then actual IPC speed...That is well above the 20% conroe was leading in real world apps that often dont count on the cache as much....

If you look at it the P4's and the netburst had done quite well in superpi over the last few years and we know where ther were landing in real world apps....

So in conclusion once those guys stop hand jacking themselves over worthless benches and start running real world apps ...LET ME KNOW

Exactly !

 

[GuitarDaddy]

I see he stopped the 32m run because of bios problems or at least he's not publishing the results. I'm guessing he didn't like the numbers he was seeing.

I agree SuperPI is a pretty crappy CPU test, I wish they would run PCmark or Crystalmark or some encoding benches.

I see all the guys at XS are screaming for 3dmock and gaming benches. Jeez
Thats like buying a vette and the first test you want to run is to see how many groceries you can carry in it.

 

[JSFLY]

Originally posted by: GuitarDaddy
I see he stopped the 32m run because of bios problems or at least he's not publishing the results. I'm guessing he didn't like the numbers he was seeing.

I agree SuperPI is a pretty crappy CPU test, I wish they would run PCmark or Crystalmark or some encoding benches.

I see all the guys at XS are screaming for 3dmock and gaming benches. Jeez
Thats like buying a vette and the first test you want to run is to see how many groceries you can carry in it.

lol good analogy

Imo, the conroe is overhyped. I will make my decision on what I think about it after it comes out.

 

[nyker96]

I can't say I'd equate real world gain with bench gain. A 100% gain on spi isn't going to effect your real world by nearly that much, however, considering it's only running at 2.4 Ghz and is rumored to ramp up to 3.5+, this could be a formidable CPU when it become mass produced. I have also read a bunch on gaming benches, also seems quite good for their speed. Excellent result in all but don't take that spi gain = real world tho.

 

[n19htmare]

Doesnt the 9xx series have 4MB of L2? I don't see the 9xx series doing Low 20s in Sp1.....
Do you think it might have to do with the actual architect of the CPU itself....perhaps the shorter pipelines....

 

[BrownTown]

well, no crap the shorter pipeline has alot to do with it, but the Conore has 4MB of cache for SuperPi, and 900 series, dothan, yonah, and Cedar Mill all have only 2

 

[n19htmare]

Well since the Conroe uses all 4MB of L2 for single threaded apps AND it has shorter pipelines.... AND from the early previews/reviews/benches...It's kinda hard to miss the fact that it's a great processor.
seems to me that at $310 vs the $1000+ price tag of FX60...the choice is simple.

 

[Tangerines]

Wow! And to think, this is a middle-of-the-line, non-overclocked Conroe. I can't wait to see what the higher end models do.

 

[Elfear]

Originally posted by: GuitarDaddy
I see he stopped the 32m run because of bios problems or at least he's not publishing the results. I'm guessing he didn't like the numbers he was seeing.

I agree SuperPI is a pretty crappy CPU test, I wish they would run PCmark or Crystalmark or some encoding benches.

I see all the guys at XS are screaming for 3dmock and gaming benches. Jeez
Thats like buying a vette and the first test you want to run is to see how many groceries you can carry in it.

Personally, I like to see the gaming and benching results because that's what I do with my rig. I know a lot of guys encode and use their computers for different things, but a good share of us put a lot of stock in gaming potential when we evaluate a processor.

I'd tweak your analogy a bit and say it's like buying a new Vette and seeing how she does at the dragstrip. That's why I'd buy a Vette anyway, to see how fast she is. Everyday commuter qualities are important, but benchmarks are more interesting.

 

[spittledip]

Whenever I see benchmarking programs like PCmark or whatever, I just ignore them b/c I am only interested in what the hardware can do with real world apps.

 

[Hard Ball]

Originally posted by: RichUK

Originally posted by: Furen
All I have to say is 4MB L2. Going to 32M would require the chip to hit the ram which would affect performance tremendously.

Exactly :thumbsup:

I would have to say that at this point, SuperPI at 1M is meaningless. We've always know that Dothans did much better on superPI than comparable chips, because of the access to memory is restricted to L2. Loading the register file with data cached in L2 is typically 10X to 30X lower in latency than accessing system memory.

SuperPI even at 2M with 4MB of L2 of the conroe is bringing similarly meaningless result. I was hoping that xtremesystems would give us something a little more meaty, but it's looking like another uninformed website doing some inconsequential bench that will not really tell us anything.

edit: typos

 

[Elfear]

Originally posted by: Hard Ball

SuperPI 2M with 4MB of L2 of the conroe is bringing similarly meaningless result. I was hoping that xtremesystems would give us something a little more meaty, but it's looking like another uninformed website doing some inconsequential bench that will not really tell us anything.

Woah. Give the guys over there a break. Benchmarking is what they like to do so of course those are the first results you're going to see. Should they call you uninformed because you don't like to bench? Are the things they use their computers for inconsequential because YOU don't find them useful? Different strokes for different folks.

Just be patient. I'm sure they'll run more benchmarks than just SuperPi. Sheesh.

 

[Markfw]

Originally posted by: Tangerines
Wow! And to think, this is a middle-of-the-line, non-overclocked Conroe. I can't wait to see what the higher end models do.

I thought that 2.13 was the fastest, and it was over 530 ? He OC'ed a $530 chip I think, right ?

Edit, just saw the revised pricing. But I still want to see what the mobo cost, and the real thing at launch.

 

[Elfear]

Originally posted by: Markfw900

I thought that 2.13 was the fastest, and it was over 530 ? He OC'ed a $530 chip I think, right ?

Here is a pricing list with most of the future chips and their clock speeds shown. Heard there was a 3.33GHz EE version coming .:Q

 

[Furen]

A superPi 1m run took him 20 seconds... a 32m should take less than half an hour yet he only did the 20 second one. Yes, I know it's impressive but only doing SuperPi 1M on the CPU everyone is looking forwards to either implies that he didnt like the answers and decided to look into them (kind of like anandtech did with socket AM2) or that he's just being a teasing moron.

Let me rephrase the cache thing... It has 4MB of low-latency L2 on a highly efficient processor design. If it hits the main memory it should still take a pretty big performance hit.

 

[Duvie]

Originally posted by: Elfear

Originally posted by: Markfw900

I thought that 2.13 was the fastest, and it was over 530 ? He OC'ed a $530 chip I think, right ?

Here is a pricing list with most of the future chips and their clock speeds shown. Heard there was a 3.33GHz EE version coming .:Q

how comes it only shows the cache at 2mb??? I know Fugger has an ES sample....

 

[Elfear]

Originally posted by: Duvie

how comes it only shows the cache at 2mb??? I know Fugger has an ES sample....

Good question. Everyone over at XS is saying 2x2MB??