Sign in with
Sign up | Sign in
Your question

AT conroe review updated

Tags:
Last response: in CPUs
Share
March 9, 2006 10:02:29 AM

Quote:
http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716...
Still,it doesn't change much. Conroe is way ahead.


Because of DDR2. If you look at straight CPU tests (Encoding) Conroe is barely ahead, which shows that Conroe is barely faster given it's new Architecture, that is a fact, not a fanboy comment. I still want to know what this "Modified ATI Driver" did and if it could have affected the AMD System, which it could have very easily.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 9, 2006 10:09:03 AM

So DDR2 will give amd a 20% performance increase? Wow imagine what DDR3 will do....
Related resources
March 9, 2006 10:10:31 AM

Quote:
So DDR2 will give amd a 20% performance increase? Wow imagine what DDR3 will do....


Conroe does not have a 20% increase in CPU power, those Gaming Benchmarks are RAM related, so yes, DDR2 will give the performance crown in Gaming to a Par or to either Conroe or AM2 by <5%. But those encoding Benchmarks prove the CPU power in Conroe is not >20%.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 9, 2006 10:10:48 AM

DDR2 can't be the ONLY cause now can it?
Anyway,looking forward to see how DDR2 affects AMD's performance then we'll have a clear picture.
March 9, 2006 10:12:29 AM

Quote:
DDR2 can't be the ONLY cause now can it?
Anyway,looking forward to see how DDR2 affects AMD's performance then we'll have a clear picture.


Read my previous post, DDR2 will bring only gains in Gaming, where RAM Bandwidth is very important.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 9, 2006 10:15:47 AM

After AnandTech reviewed their tests:

Quote:
The performance picture with regards to Conroe hasn’t really changed all that much - on average we’re still seeing a bit over a 20% increase in performance over an overclocked Athlon 64 FX-60.
March 9, 2006 10:17:54 AM

Quote:
After AnandTech reviewed their tests:

The performance picture with regards to Conroe hasn’t really changed all that much - on average we’re still seeing a bit over a 20% increase in performance over an overclocked Athlon 64 FX-60.


That's because all they looked at were Gaming Benchmarks where the DDR2 gave Conroe the advantage, look at the Encoding, they are 10%/12%/30% gain over FX-60. DiVx can use a tremendous amount of Memory which can quickly give the 30% gain attribution to the DDR2 vs. DDR1, BTW, in no way do I believe it was OC'd to 2.8 correctly, wouldn't doubt if the FSB is set to 100MHz.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 9, 2006 10:29:41 AM

The thing that get me is that this chip is 6 months away from being released. Why? If it's this good, why wait. Are they still tweaking it? My guess is that they are filling up the pipeline. Oh yeah and by the way, has anyone heard of anymore dell going to amd? I guess sooodoo or whatever his name was way off.
March 9, 2006 10:29:56 AM

Quote:
That's because all they looked at were Gaming Benchmarks where the DDR2 gave Conroe the advantage, look at the Encoding, they are 10%/12%/30% gain over FX-60. DiVx can use a tremendous amount of Memory which can quickly give the 30% gain attribution to the DDR2 vs. DDR1, BTW, in no way do I believe it was OC'd to 2.8 correctly, wouldn't doubt if the FSB is set to 100MHz.


But isn´t gaming performance what really made it for Athlon 64?
One more thing, AT states that they´ve checked out all BIOS settings (they actually instaled the last BIOS version), drivers and stuff to see if anything might be wrong. Even thought they might have not done their first reiew correctly, they have gone back and checked it all. And at the end they got results very similar to the ones before, and some even better for Conroe.
Anyway, we should wait for the release, so we can truly see what´s coming from Intel (seems there´s a 3 GHz Conroe) and AMD AM2 plataform, which the latest news I´ve seen around have shown worse performance than the 939 (AMD is having some hard time with the DDR2 HT stuff).


PS - Don´t think I´m Intel Fanboy, cause I haven´t had an Intel processor since Pentium 3 (and that´s over 7 years ago).

Edit: typo on the text :p 
March 9, 2006 10:32:50 AM

Quote:
That's because all they looked at were Gaming Benchmarks where the DDR2 gave Conroe the advantage, look at the Encoding, they are 10%/12%/30% gain over FX-60. DiVx can use a tremendous amount of Memory which can quickly give the 30% gain attribution to the DDR2 vs. DDR1, BTW, in no way do I believe it was OC'd to 2.8 correctly, wouldn't doubt if the FSB is set to 100MHz.


But isn´t gaming performance what really made it for Athlon 64?
One more thing, AT states that they´ve checked out all BIOS settings (they actually instaled the last BIOS version), drivers and stuff to see if anything might be wrong. Even thought they might have not done their first reiew correctly, they have gone back and checked it all. And at the end they got results very similar to the ones before, and some even better for Conroe.
Anyway, we should wait for the release, so we can truly see what´s coming from Intel (seems there´s a 3 GHz Conroe) and AMD AM2 plataform, which the latest news I´ve seen around have shown worse performance than the 939 (AMD is having some hard time with the DDR2 HT stuff).


PS - on´t think I´m Intel Fanboy, cause I haven´t had an Intel processor since Pentium 3 (and that´s over 7 years ago).

Conroe now has an architecture more similar to the A64 and now DDR2 is being able to take effect (800MHz FSB can't even handle fast DDR2) and you're seeing the gain DDR2 gets in gaming, look at the Encoding, where its CPU vs. CPU, barely a gain given by Conroe over K8.

BTW: I think you just made up something else, there is no problem w/ DDR2+HyperTransport.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
a b à CPUs
March 9, 2006 11:05:39 AM

Quote:
That's because all they looked at were Gaming Benchmarks where the DDR2 gave Conroe the advantage, look at the Encoding, they are 10%/12%/30% gain over FX-60. DiVx can use a tremendous amount of Memory which can quickly give the 30% gain attribution to the DDR2 vs. DDR1, BTW, in no way do I believe it was OC'd to 2.8 correctly, wouldn't doubt if the FSB is set to 100MHz.


But isn´t gaming performance what really made it for Athlon 64?
One more thing, AT states that they´ve checked out all BIOS settings (they actually instaled the last BIOS version), drivers and stuff to see if anything might be wrong. Even thought they might have not done their first reiew correctly, they have gone back and checked it all. And at the end they got results very similar to the ones before, and some even better for Conroe.
Anyway, we should wait for the release, so we can truly see what´s coming from Intel (seems there´s a 3 GHz Conroe) and AMD AM2 plataform, which the latest news I´ve seen around have shown worse performance than the 939 (AMD is having some hard time with the DDR2 HT stuff).


PS - on´t think I´m Intel Fanboy, cause I haven´t had an Intel processor since Pentium 3 (and that´s over 7 years ago).

Conroe now has an architecture more similar to the A64 and now DDR2 is being able to take effect (800MHz FSB can't even handle fast DDR2) and you're seeing the gain DDR2 gets in gaming, look at the Encoding, where its CPU vs. CPU, barely a gain given by Conroe over K8.

BTW: I think you just made up something else, there is no problem w/ DDR2+HyperTransport.

~~Mad Mod Mike, pimpin' the world 1 rig at a time

P4's were using DDR2 and they wernt fast, and AM2 with the same speed ddr2 (667) didnt boost any (5% around, better and worse at the benchies)

Conroes architecture similar to A64? thats total BS - its based on good old P6 which was designed in 95, and AMD copied the concepts used to make the K6 (notice the 6?) which advanced to the K7 and K8 today - Conroe has nothing to do with the A64.
March 9, 2006 11:12:13 AM

Where did k6 have onboard memory controler? Where did k6 have hypertransport. Hum I dont see any.

The only chip that I see that uses p3 tech. Is Laptops. Intel P M chips.
March 9, 2006 11:16:17 AM

Well, I´ve read in a few places about the bandwith problem AM2 was getting...

Check out this link (it´s from this week):

http://www.anandtech.com/weblog/default.aspx

But I do expect AMD to change this. They still have 3 months or so before launch. Althought I might not be making any major upgrade for a long time (maybe I´ll put some more RAM or go dual core on the 939, but not anything as drastic as changing to Conroe or AM2).
a b à CPUs
March 9, 2006 11:25:57 AM

Quote:
Where did k6 have onboard memory controler? Where did k6 have hypertransport. Hum I dont see any.

The only chip that I see that uses p3 tech. Is Laptops. Intel P M chips.


Huh? what do you mean?

Amd for years has improved there design, intel also should have but then we got P4, and now were back to the P6 design (from pentium m which is based on P6).

P6 tech evolved to the P2, P3, then Pentium M and now to conroe.

The original pentium pro (P6) shared the same chipset as the P2, and a P2 could fit the P3 chipsets etc - they were the same design with a few updates, Pentium M changed the FSB so it was incompatible but otherwise it was a tualatin with a P4 fsb and twice the cache.
March 9, 2006 11:56:59 AM

Shhhush! We do not want to hear the truth if it means AMD's AM2 using DDR2 is actually slower than with current 939 chips!
That is just speculation, and AMD will refine it by the time it is ready for mainstream to be 50% faster than any other chip, clock for clock!
Nor do we want to hear that Intel's Conroe preview actually performs better than todays faster current 939 chip even if they released a faster FX chip than currently available, for by actual release day they will have slowed down the chip significantly to nestle in right under the performance levels of AMD soon to be released chips....
All is well in the world now... ;) 
March 9, 2006 12:00:24 PM

what is crazy is that they are going against a overclocked FX-60. and still the preformance....

Now i konw that wastnt the extreme edition of the Conroe...

So AMD has to release something.. Something tight.
I cant wait for the top vs top test to hit...
March 9, 2006 12:10:47 PM

Quote:
Where did k6 have onboard memory controler? Where did k6 have hypertransport. Hum I dont see any.

The only chip that I see that uses p3 tech. Is Laptops. Intel P M chips.


You do realize that all of Intels new processors are based on the Pentium M right? And the Pentium M is based on the Pentium III.

I see the arguement mike is trying to make. But sorry Mike, I don't see DDR2 giving AMD the huge boost in performance you do. THG did tests with an early AM2 chip. Granted it was with DDR2 667 instead of DDR2 800 but I don't see there being that huge of a difference between the two of them.

Once again though the best idea is for all of us to just shut up and wait.
March 9, 2006 12:15:02 PM

ok guys im back so whats going on today *looking at replies* oh no not this crap agian huh :roll:
March 9, 2006 12:22:24 PM

Quote:
I see the arguement mike is trying to make. But sorry Mike, I don't see DDR2 giving AMD the huge boost in performance you do. THG did tests with an early AM2 chip. Granted it was with DDR2 667 instead of DDR2 800 but I don't see there being that huge of a difference between the two of them.


That was becasue that chip had the DDR2 bug. That's the reason AMD didn't want anyone to preview the chip, since it will give a false impression of the chip's performance.

Conroe is going to be the best cpu that intel has ever come out with, but amd will be right there with it.

intel bringing the conroe out now was to knock the wind out of amd's sales , but what intel forgot was that thier sword is a double edge sword, and the blade fell back and has killed off intel sales

yes the 2.6 conroe looks good , but this is the best conroe that intel was able to bring to the show

there is a difference between a hand picked cpu , and putting out millions of them

i hope intel can do it , the competition is good for all of us.

Let's not forget what the FX-62 will look like on the AM2 platform. I'm sure AMD knew that all of this was going to happen (in some kind of way). This will give them enough time to tweak the FX-62 for it to battle against Intel's conroe EE version. :wink:
March 9, 2006 12:24:41 PM

Quote:
those Gaming Benchmarks are RAM related

That would put the FX way ahead :roll:
March 9, 2006 2:18:12 PM

Quote:
http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716...
Still,it doesn't change much. Conroe is way ahead.

but F.E.A.R's benchmark actually decreased from 36% advantage to 19% advantage, and you're telling me it doesn't change much?
ok.. maybe the bios problem we caught this time only reduces their nearly 40 % increase to 20% increase in performance.

however, intel is known to do anything, and i mean ANYTHING to destroy the credibility of the benchmark.
history:
intel is known to have used CPUID to reduce performance in other machines in benchmarks and programs, such as BAPco, Skype, and a linux compiler.
http://www.swallowtail.org/naughty-intel.html
what does this tell us? intel does not have the credibility to be trusted with building their and rival's machine on their own. AT only said they found no problem with hardware configuration, but never mentioned the software benchmarks they were using. would anybody know if there are CPUID dirty tricks within their code segments?

such a big coporation, but intel has to rely on using little dirty tricks to compete with its rival. what a good model for us to learn from.
March 9, 2006 2:22:55 PM

Intel isnt in ANY need to do that plus why would they want to do that considering that in 3 months any hardware site could prove them wrong?
March 9, 2006 2:27:55 PM

Quote:
Intel isnt in ANY need to do that plus why would they want to do that considering that in 3 months any hardware site could prove them wrong?


Intel can do this because some of us will halt the purchase to wait for Cornoe.

Also from the Prescott benchmarks, Intel had done this before.
March 9, 2006 2:30:35 PM

oh yeh... AT is also known for performing unfair benchmarks w/ server cpus.
http://www.anandtech.com/IT/showdoc.aspx?i=2644&p=3
they did 32bit test on both opty and xeon, when they knew opty performs better in 64bit condition. in this situation, it clearly favors int-hell, who doesn't really know how to incorporate 64 bit tech with dual core, and only has advantages in 32bit.

ooo wait.. before i forget, AT did 32bit tests, so that means disadvantages for AMD because AMD only incoporated 32bit as "complement measures". does this mean opty is worse than xeon? it only means xeon performs better than opty under 32 bit. but as more companies use 64 bit servers, AT test would be inconclusive, or to state the fact, opty performs better than xeon in 64bit, which is the trend for the future.

anand tech + int-hell = the best lying team of all times
March 9, 2006 2:32:48 PM

Quote:
but F.E.A.R's benchmark actually decreased from 36% advantage to 19% advantage, and you're telling me it doesn't change much?
ok.. maybe the bios problem we caught this time only reduces their nearly 40 % increase to 20% increase in performance.

however, intel is known to do anything, and i mean ANYTHING to destroy the credibility of the benchmark.
history:
intel is known to have used CPUID to reduce performance in other machines in benchmarks and programs, such as BAPco, Skype, and a linux compiler.
http://www.swallowtail.org/naughty-intel.html
what does this tell us? intel does not have the credibility to be trusted with building their and rival's machine on their own. AT only said they found no problem with hardware configuration, but never mentioned the software benchmarks they were using. would anybody know if there are CPUID dirty tricks within their code segments?

such a big coporation, but intel has to rely on using little dirty tricks to compete with its rival. what a good model for us to learn from.


If you read the update AT did on the review, you´ll see that they installed their own version of the benchmark programs...So they were not tweaked in any matter by Intel.

Btw, I´m not defending Intel, but they did show some good piece of hardware. Intel is like any other company they give a crap about us and only focus on getting their profit´s higher as they can, no matter what they have to do, after all, it´s Machiavel - The ending justifies the means (i.e., American counter-terrorism actions: do evil to end an even bigger evil). :evil: 
March 9, 2006 2:32:56 PM

Quote:
Intel isnt in ANY need to do that plus why would they want to do that considering that in 3 months any hardware site could prove them wrong?


Intel can do this because some of us will halt the purchase to wait for Cornoe.

Also from the Prescott benchmarks, Intel had done this before.
that's why i think intel has a better marketing strategy... by hyping the enthusiast markets with a 6 months away processor, with benchmarks that's nearly inconclusive.

oh yeh.. intel claimed that conroe has 40% decrease in power consumption. where is the fact for that? maybe they were running dual liquid cooler within the conroe case.
March 9, 2006 2:38:08 PM

That would increase power consumption *lol*

YOU are talking about hype? YOU? I bet you are one of those no-brainers that are still hoping DDR2 will put AMD on top and YOU are talking about hype?
So funny :lol:  :lol: 
March 9, 2006 2:40:42 PM

Quote:
That would increase power consumption *lol*

YOU are talking about hype? YOU? I bet you are one of those no-brainers that are still hoping DDR2 will put AMD on top and YOU are talking about hype?
So funny :lol:  :lol: 


Do you know the performance of a RETAIL Conroe and Windsor?

If not, please shut up.
March 9, 2006 2:41:55 PM

Quote:
The thing that get me is that this chip is 6 months away from being released. Why? If it's this good, why wait. Are they still tweaking it? My guess is that they are filling up the pipeline. Oh yeah and by the way, has anyone heard of anymore dell going to amd? I guess sooodoo or whatever his name was way off.


it better this way than doing what AMD did by prematurly releaseing products and withdrawing supplies abrubtly without due explantions. Dubious and criminal if you ask me.

secondly, Anandtech is known for its objective hardware reviews, i dont see any of you "wiseguys" opening your own review website. i dont need to be an Einstein to know that they have more qualified experts than most of you. One of you even went ahead claiming that the overclock done to the AMD benchmark was rigged 8O

As i have said before, wait until there are other harware reviews that prove otherwise, until then do yourself (and us ) a favour curbing all that oral diarrohea :!: :x :x
March 9, 2006 2:43:51 PM

Yea, I know Conroe´s and I know AM2 processor´s performance (which is pretty poor compared to S939).
And Im not the only one to know that, the whole world does but you probably just cant see it through your green eyes.
March 9, 2006 2:45:35 PM

Quote:
Yea, I know Conroe´s and I know AM2 processor´s performance (which is pretty poor compared to S939).
And Im not the only one to know that, the whole world does but you probably just cant see it through your green eyes.


I REPEAT, DO YOU HAVE A RETAIL Conroe and Windsor?
March 9, 2006 3:01:46 PM

Intel are obviously very confident about releasing the benchmarks so early, so a 20% lead may well hold true against current AMD CPUs (even overclocked ones) (I would hope so too, given that it is a new architecture with double the cache).
Toms AM2 review showed that even with a MC bug and slow DDR the AM2 was roughly on par with the current A64 chips.
So lets assume intel did fudge their own results a little (lets say by 5%)
and the AM2 may come out at 5% faster than the current A64 (at the same clock), this still gives the new intel chip a 10%-15% performance advantage.
Sounds roughly like the advantage AMD had over intel when they first released the A64. (if I can recall correctly)

However I personally think Intel have released these benchmarks far too early, this does give AMD a little time to tweak the next release to close the gap a little now intel has shown their ace card.

What I can see happening is that AMD may be forced into ramping up clockspeeds, once they have finished transitioning to 65nm (which they may have to acclerate) to keep up with intel, until they can release their next gen CPU.
Interesting times are ahead, and that can only be good for us.
March 9, 2006 3:13:32 PM

Quote:


However I personally think Intel have released these benchmarks far too early, this does give AMD a little time to tweak the next release to close the gap a little now intel has shown their ace card.

What I can see happening is that AMD may be forced into ramping up clockspeeds, once they have finished transitioning to 65nm (which they may have to acclerate) to keep up with intel, until they can release their next gen CPU.
Interesting times are ahead, and that can only be good for us.


Very professional put Stimpy :wink: INTEL and AMD zealotry is just plain bull***t and we ought to look at stuff with your kind of profesionalism.


Kudos.
March 9, 2006 6:34:58 PM

In the server space this new architecture could actaully kill their Itanium range.. In most server benchmarks (Business rather FP intensive apps)
the Opteron is already within 10-20% of Itanium results on a per core basis (on a per socket basis Opteron wins hand down).
Now if woodcrest is indeed 20% faster that will eat further into their itanium sales, unless they cripple woodcrest such that it can't work in more than 2/4 socket variations (i.e. upto 16 way with the release of quad core).
Woodcrest/Opteron benches will be interesting though when it comes to true SMP apps with 4 CPUs/(or cores) and above, as this is where the onboard MC really shines.

It makes my job harder now though, as I have to take yet another architecture into consideration when forcasting my companies future IT needs.
(we have just got 200 HP Dual socket Dual core Opteron blades running VMware, to replace our entire citrix estate, thank god)
March 9, 2006 7:28:41 PM

lol.

This is bad, but I strangely can't resist posting.

To anyone who believes moving from DDR to DDR2 alone will improve performance: I have a muffler for your car I can sell you that will give if 100 horsepower!

Seriously, though, DDR to DDR2 increases bandwidth AND latency. I don't care if you like AMD or intel, facts are facts. The ARCHITECTURE of the memory controller has to be more tolerant of these latencies.

I'm sure AMD is making more changes than DDR2 to thier AM2 lineup.
March 9, 2006 7:30:26 PM

Intel will probably be retiring the Itanic...
March 9, 2006 7:30:52 PM

I have a apple corer aluminum device that you simply attach to your carb and double your mileage and increase HP by 25%! :lol: 
March 9, 2006 7:34:14 PM

ya its called an exhuast
March 9, 2006 7:34:17 PM

One thing that must be realized ... and I say this from a completely neutral stand point between the two companies ... When AMD puts out their 65nm series on the M2 with DDR2 working with it I have a funny feeling that things will end up back the way they are now.
March 9, 2006 7:35:42 PM

you mean when amd owned prescott?
March 9, 2006 7:35:52 PM

Then you shouldnt trust your feelings.
March 9, 2006 7:37:24 PM

you should stop posting your bull crap! :evil: 
March 9, 2006 7:46:49 PM

With a bigger cache (Which AM2 promises to have) and some tweaks to the memory controller we'll definately see some improvements in gaming performance and memory-intensive server apps.

I doubt that increase will be 20%, but I think the AM2 might be going for the marketing at this point over bleeding-edge performance. Marketing in terms of Performance-Per-Watt. The only market AMD can't seem to penetrate si the mobile market, depsite the fact that they've show consistent superiority in low-power.

I don't think AMD will beat Conroe in raw performance, but performance per watt. PPW will give them an edge in the mobile market as well as allowing them to market PPW to the system builders of the world.

Everyone loves the enthusiest, but admit it, we're a small part of the market. I don't think letting Intel have a 5% performance edge on AM2 will have AMD worried, especially if they can further lower power consumtion.
March 9, 2006 8:27:58 PM

I think Intel will have Amd beat on performance per watt, since the conroe that was demoed was clocked at 2.6 and the fx was at 2.8. Raw performance wise they will be able to compete with intel, but will have to sacrifice higher wattage since they would probably have to clock up > 3.0ghz to keep up. That's until amd get 65nm going and then we'll see what happens.
March 9, 2006 8:29:07 PM

Eat sh!t fanboys. Wrong again.
March 9, 2006 8:32:41 PM

Did I mis-interpret the specs on Conroe being lower power per watt than AMD?
AMD has proven lower power chips in the mobile market???

Got any links to back that up?
March 9, 2006 10:56:23 PM

Quote:
I doubt that increase will be 20%, but I think the AM2 might be going for the marketing at this point over bleeding-edge performance. Marketing in terms of Performance-Per-Watt. The only market AMD can't seem to penetrate si the mobile market, depsite the fact that they've show consistent superiority in low-power.


Unfortunately this is not true. In the current desktop market AMDs consume significantly less power than Intels. However in the mobile market the Intel chips perform better and consume less power. AMD is planning a dual-core version of their Turion but I am willing to bet it will use more power and perform worse than Yonah, let alone Merom which will likely be released shortly after the Turion 64 X2's.

The entire lineup of new Intel CPUs was derived off of the mobile architecture in the order P6 -> Banias -> Dothan -> Yonah -> Merom. Conroe and Woodcrest, and their derivatives, are modified versions of the original Merom design.

Back before Intel started hyping up the performance increase with their new microarchitecture, the most important thing they were marketing was low power usage and high performance per watt (PPW).

Honestly, at that time (about 6 months ago), I believe that Intel did not know if they would beat K8 architecture. I think they were afraid that they would not and so they marketed an increase in performance over their own current architecture (Netburst) in addition to their low power initiative.

Recently, Intel likely finished designing their hardware and created manufacturing samples for testing and were surprised by the jump in performance which is why they marketed the chips differently at this IDF--much like the cat that got the cream, they can't stop licking themselves.
March 9, 2006 11:01:25 PM

Merom is not based off Yonah and conroe and woodcrest arent derived from merom. Merom, Conroe and Woodcrest are all from the same family.
March 9, 2006 11:04:03 PM

[quote="anand tech + int-hell = the best lying team of all times[/quote] True or not, I never really liked either.
!