Sign in with
Sign up | Sign in
Your question

AMD 65nm in June

Last response: in CPUs
Share

Which chip will be the first 65nm from Charter?

Total: 23 votes

  • Turion/Turion X2
  • 18 %
  • Athon 64/FX X2
  • 83 %
March 31, 2006 6:47:58 PM

As I thought AMD is releasing SOME 65nm chips this year. It seems that the XBox chip manuf has been called upon by AMD to second source their manuf. It isn't known what chip famiy the first chips will represent but my money is on Turion X2.

AMD will own games and quad servers for the foreseeable future so they should focus on beatign Core Duo - plus 64bit dual core.

The news is over at AMDZone.com. Any guesses as to hwih family the first chips will belong to? An FX62 would be nice or maybe a 890 Opteron, but they don't need another boost there before Conroe.

2.2 GHz Turion X2 anybody?

More about : amd 65nm june

March 31, 2006 6:53:56 PM

Quote:
As I thought AMD is releasing SOME 65nm chips this year. It seems that the XBox chip manuf has been called upon by AMD to second source their manuf. It isn't known what chip famiy the first chips will represent but my money is on Turion X2.

AMD will own games and quad servers for the foreseeable future so they should focus on beatign Core Duo - plus 64bit dual core.

The news is over at AMDZone.com. Any guesses as to hwih family the first chips will belong to? An FX62 would be nice or maybe a 890 Opteron, but they don't need another boost there before Conroe.

2.2 GHz Turion X2 anybody?


I can almost guarantee you the first AMD64 CPU's to receive the 65nm upgrade will be the Server Opteron 64's.

I vote C: Opteron 64

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 6:57:50 PM

Opterons on 65 are the only chips with a possibility of appearing this year in Q4 2006
AM2 +Turion 65 nm will arrive in Q1/2 2007.
Related resources
March 31, 2006 6:59:36 PM

I kinda thought that at first but realized that AMD doesn't need to do anything in the server space. They kick Intels butt in Apache AND SQL. Even Conroe/Woodcrest won't hurt them in the 4Way space. They do need to make a more compelling case for the mobile space though.

Thx for the comments though.
March 31, 2006 7:01:41 PM

Quote:
Opterons on 65 are the only chips with a possibility of appearing this year in Q4 2006
AM2 +Turion 65 nm will arrive in Q1/2 2007.


What makes you say that? Like I said they already own the server space so why not help out the mobile business to confront Core Duo on Windows and Mac?
March 31, 2006 7:12:11 PM

Quote:
Opterons on 65 are the only chips with a possibility of appearing this year in Q4 2006
AM2 +Turion 65 nm will arrive in Q1/2 2007.


AMD is concentrating on the Server and Mobile Market now, and there is a GUARANTEE of 65nm this year, but the platform of choice is unknown, but as I said, more likely Opteron 64's.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:14:00 PM

A guarantee? Really... AMD guarantees it???
March 31, 2006 7:16:10 PM

During a LapDance at the local strip joint... I can guarantee something that will come out of it...

But AMD is like that too??? :wink:
March 31, 2006 7:18:04 PM

Quote:
A guarantee? Really... AMD guarantees it???


As guaranteed as I am to finish my Arby's I'm eating right now and say "Damn...that's good sh!t!!" when I finish.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:18:32 PM

Y i know AMD is trying to hold the server lead(that is the real jackpott to fast growth). Opterons on 65 nm sounds promising. Will they need ECC ram? And will there be cheaper variants like with the current opteron.
65 nm should give it more headroom and it will be intresting to see some OCing results.
March 31, 2006 7:20:07 PM

Quote:
Y i know AMD is trying to hold the server lead(that is the real jackpott to fast growth). Opterons on 65 nm sounds promising. Will they use ECC ram? And will there be cheaper variants like with the current opteron.
65 nm should give it more headroom and it will be intresting to see some OCing results.


What I want is a 1xx 65nm Opteron on Socket AM2. I can get a 90nm Opty 1xx to over 300MHz FSB @ stock voltage, imagine 65nm....

DAMN THAT WAS GOOD SH!T! - See RichPLS, guaranteed ;) .

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:25:57 PM

Quote:
A guarantee? Really... AMD guarantees it???



I guess the same way they say the Turion X2 WILL com out on May 9th. Teh word guarantee doesn't really fit int his case. Scheduled is a better word.

Like AMD scheduled the release for June 2006. Hey isn't that hen AM2 comes out. Hmmm, I remember saying that AMD had a few tricks and COULD or probably WOULD release 65nm chips this year. They maybe doing it to get the ZRam and L3 out sooner rather then later. Maybe it will also help to clock the RAM to DDR2 1066.

I still bet on the Turion X2, maybe with some low power or higher clocked Opterons going to Sun.
March 31, 2006 7:28:50 PM

Quote:
Y i know AMD is trying to hold the server lead(that is the real jackpott to fast growth). Opterons on 65 nm sounds promising. Will they need ECC ram? And will there be cheaper variants like with the current opteron.
65 nm should give it more headroom and it will be intresting to see some OCing results.



Maybe Dell wants them for their AlienWare boxes. Maybe Sun wants them for their clusters. Maybe HP wants them fo rtheir DTRs. I would bet STILL on the mobile chips. MAYBE the fastest Opterons, MAYBE.

But I guess those of us who are still alive will find out in June.
March 31, 2006 7:28:56 PM

I think Booh!rion because its power consumption is closest to Core Duo so it might have a more noticable effect if Booh!rion is the first on 65nm
March 31, 2006 7:30:41 PM

Quote:
I think Booh!rion because its power consumption is closest to Core Duo so it might have a more noticable effect if Booh!rion is the first on 65nm


Since a Turion has lower power consumption than Merom, not sure why you would need 65nm...it'd just be a slap in the face to Merom.

~~Mad Mod Mike, pimpin' the world 1 rig at a time
March 31, 2006 7:35:20 PM

Quote:
I think Booh!rion because its power consumption is closest to Core Duo so it might have a more noticable effect if Booh!rion is the first on 65nm


Exactly, P4 EEs are not a threat and Xeon EEs are not a threat. Only the Core Duo is outshining an AMD part and ONLY for power. I'm sure a dual Turion will kick Intel's butt.
March 31, 2006 8:24:09 PM

Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Poorion in almost every situation. The same will go for Dual Ms and dual Poorions.
March 31, 2006 8:43:32 PM

I think that we will not see 65nm K8 chips this year... :( 
March 31, 2006 10:07:11 PM

Quote:
I kinda thought that at first but realized that AMD doesn't need to do anything in the server space. They kick Intels butt in Apache AND SQL. Even Conroe/Woodcrest won't hurt them in the 4Way space. They do need to make a more compelling case for the mobile space though.

Thx for the comments though.


You must have missed the Intel Xeon LV edition (based on 1st gen Core Duo) in mid March this year (2006).

Plug-in upgrade for Xeon server based mainboards, higher performance and far lower power usage, and heat output. (It is no Woodcrest, but it'll do for now).

AMD need to compete in the server space still.

My money is on a drop in 65nm replacement for Opteron 200 to 800 series, on Socket 940, to compete with the Xeon LV, and continue competing even when (Inte server processors based on) Woodcrest come out. They have an excellent chance here, need the extra production capacity and should seize it.

Maybe even a few 65nm in the Opteron 100 series for both Socket 939 & 940 for those more 'cost effective' server systems (eg: servers on S939 w/o Registered RAM for example). They'll use less power, generate less heat, and may only require passive cooling to work, although they'll get active cooling regardless.

If it is enough for datacentres not to justify the cost to migrate 'back' to Intel they'll be happy. AMD don't want to lose market share they just gained in the server area, they need to prove they are viable for 2-3 upgrades to a machine to save on overheads and maintain performance. If they jump ship back to Intel, after one 'negative' experience (eg: changing then changing back)... they'll never get those customers back again... not within 5 years anyway.

Anyways, a move like this would make MadModMike, and myself, quite happy. Rarely does one move please both myself and Mike. :p 

Even though I only aggree with MadModMike 50% of the time, this is one of those times. :p 

Quote:
Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Turion in almost every situation. The same will go for Dual Ms and dual Turions.


Yeah, except when the final machine (Turion 64) costs US$350 less, one can then upgrade it a little, buy another 2 batteries (3 total), get a dual-battery recharger, etc

For an equal cost, which one 'performs better' ?, Do bear in mind that laptops are not benchmarked by consumers using 'performance' but by battery life, quantity of RAM, HDD, etc.

I'd rather have 2 or 3 batteries (spares), more RAM and a 10% slower CPU for typing documents on the go, than a 10% faster CPU with none of the above. It costs the same, which one will consumers be advised to select ? 8) - HP/Compaq really got the Presario laptop line right, both AMD and Intel CPUs in the same line, various choices for every budget.
April 1, 2006 1:03:55 AM

Quote:
Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Poorion in almost every situation. The same will go for Dual Ms and dual Poorions.



That's not what I saw when Anand tested them. It showed the Turion on top of Yonah in a lot of situations. I'll look for the review, but even so that's more reason for them to do 65nm mobile first, don't ya think.
April 1, 2006 1:10:38 AM

Quote:
Plug-in upgrade for Xeon server based mainboards, higher performance and far lower power usage, and heat output. (It is no Woodcrest, but it'll do for now).

AMD need to compete in the server space still.

My money is on a drop in 65nm replacement for Opteron 200 to 800 series, on Socket 940, to compete with the Xeon LV, and continue competing even when (Inte server processors based on) Woodcrest come out. They have an excellent chance here, need the extra production capacity and should seize it.



You must have missed that the Quad CPU setup is OWNED by AMD. They trample the Xeon by DOUBLE in 50,000 user Apache tests and the HP 585 OWNS the TPC-H DB record up to 1TB. That means Woodcrest or Xeon LV will have to be 3-4 times faster than a Xeon out now.

I say AGAIN, AMD owns the 4 way space. Intel KNOWS they can't knock them down with a shared bus.
April 1, 2006 1:36:36 AM

I'm assuming by quad CPU you mean and 4-way you mean a 4 socket system. In that case, AMD will still have a definitive lead this year. Next year things will get more complicated with Intel releasing quad-FSBs and AMD releasing K8L. However, AMD will have a run for it's money in the 2 socket segment when Woodcrest is released.

I'll have to agree with you that the Turion X2 will probably be the first to go 65nm. The server market prides itself on stability and proven technology, so it would make sense for the 65nm process to be trialed on something else first and allow the yields to increase. With the constant rumblings of FD-SOI or improved SiGe processes, the Opteron could still get more mileage on 90nm.

Quote:
2.2 GHz Turion X2 anybody?

Besides, the Turion X2 is already set to be released at 2.2GHz although only in the 35W version with the 25W version topping out at 1.8GHz.

Quote:
I guess the same way they say the Turion X2 WILL com out on May 9th. Teh word guarantee doesn't really fit int his case. Scheduled is a better word.

I'm not sure if you heard, but supposedly the Turion X2 launch may have been delayed a month to coincide with AM2.

http://www.reghardware.co.uk/2006/03/30/amd_to_delay_tu...
April 1, 2006 1:57:13 AM

Quote:

As guaranteed as I am to finish my Arby's I'm eating right now



Arby's delivers now?... gee what will they think of next? :?
April 1, 2006 3:33:41 AM

Quote:

You must have missed that the Quad CPU setup is OWNED by AMD. They trample the Xeon by DOUBLE in 50,000 user Apache tests and the HP 585 OWNS the TPC-H DB record up to 1TB. That means Woodcrest or Xeon LV will have to be 3-4 times faster than a Xeon out now.

Of course, IBM's 4S Xeon MP systems wins the 1TB benchmark for TPC-H. And it's also the fastest 4S system for TPC-C, which is the more important benchmark. And it's also fastest for 8S and 16S, but then Opterons can't scale efficiently past 4S anyways with the current glueless infrastructure, unlike the IBM Xseries which can go up to 32S.
April 1, 2006 3:56:59 AM

Hey Madmodmike, try this:
Get a turion 64 and a Intel Pentium M 1.6 and the Microsoft Excel Macro program to analyze stock indicators. Run it the same time. It will tell you who the true winner is. Dare you. It will open your eye's. 8O
April 1, 2006 7:55:03 AM

Quote:
Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Poorion in almost every situation. The same will go for Dual Ms and dual Poorions.



That's not what I saw when Anand tested them. It showed the Turion on top of Yonah in a lot of situations. I'll look for the review, but even so that's more reason for them to do 65nm mobile first, don't ya think.
Links are your friends. The only thing Anand compared was a Core Duo against a X2.
April 2, 2006 1:08:41 AM

Quote:
Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Poorion in almost every situation. The same will go for Dual Ms and dual Poorions.



That's not what I saw when Anand tested them. It showed the Turion on top of Yonah in a lot of situations. I'll look for the review, but even so that's more reason for them to do 65nm mobile first, don't ya think.

Did Anand do a direct turion to core duo compare???

I thought he compared it to an X2 3800 and X2 4000+ as a speed/cache compare.

There has been some info though on how the Turion X2 will compare to Core Duo:

Link (in German)

Hmmmmm.... is this the anand tech comparision you were refering to?

Anand Tech Yohna Part 2



Core Duo is NOT the Pentium M. I said Yonah when I should have said Dothan.
April 2, 2006 1:13:01 AM

Quote:

You must have missed that the Quad CPU setup is OWNED by AMD. They trample the Xeon by DOUBLE in 50,000 user Apache tests and the HP 585 OWNS the TPC-H DB record up to 1TB. That means Woodcrest or Xeon LV will have to be 3-4 times faster than a Xeon out now.

Of course, IBM's 4S Xeon MP systems wins the 1TB benchmark for TPC-H. And it's also the fastest 4S system for TPC-C, which is the more important benchmark. And it's also fastest for 8S and 16S, but then Opterons can't scale efficiently past 4S anyways with the current glueless infrastructure, unlike the IBM Xseries which can go up to 32S.


Above 500GB, the fastest systems are 64 proc machines. The 4 WAY SPACE IS OWNED BY AMD. The only reason Dell wins in TPC-C is because they get the best deals on Intel HW.


This link makes it clear who is the FASTEST AT 4P.

http://www.tpc.org/tpch/results/tpch_perf_results.asp

The fact that it takes CLUSTERS AT 64P to beat it shows who's running DB OLAPOLTP
April 2, 2006 1:14:17 AM

Quote:
Not as i see it. The released info about it isn't impressive.
And a Pentium M beats a Poorion in almost every situation. The same will go for Dual Ms and dual Poorions.



That's not what I saw when Anand tested them. It showed the Turion on top of Yonah in a lot of situations. I'll look for the review, but even so that's more reason for them to do 65nm mobile first, don't ya think.
Links are your friends. The only thing Anand compared was a Core Duo against a X2.


I don't look for links all the time. This is a hobby not a job. But I meant to say Dothan. That is the Pentium M, not Yonah which is? Core Duo.
April 2, 2006 1:57:02 AM

Quote:
I'm not sure if you heard, but supposedly the Turion X2 launch may have been delayed a month to coincide with AM2.

http://www.reghardware.co.uk/2006/03/30/amd_to_delay_tu...


That one slipped by, but maybe that's the reason they're pushing them back. They want to use 65nm for them. HWo knows, but funny enough the Opteron choice in the poll didn't come up.
April 2, 2006 5:14:18 AM

Quote:

That's not what I saw when Anand tested them. It showed the Turion on top of Yonah in a lot of situations. I'll look for the review, but even so that's more reason for them to do 65nm mobile first, don't ya think.


Quote:
... Did Anand do a direct turion to core duo compare???

I thought he compared it to an X2 3800 and X2 4000+ as a speed/cache ....


Quote:
Core Duo is NOT the Pentium M. I said Yonah when I should have said Dothan.


Of course it isn't, but you are not making much sense, I cannot find the Anandtech article that you refer to and therefore I can only make a guess.
Is this it??

Turion Review???

Quote:
Despite the fact that the Turion 64 line was just announced, it turns out that we actually did a Pentium M vs. Turion 64 performance comparison about a year ago. When Intel launched the Dothan Pentium M core, we compared it to the Socket-754 Athlon 64 2800+ (1.8GHz) - which is very similar to the Turion 64 ML-32 (1.8GHz/512KB). While the performance comparison isn't identical to a Turion 64 notebook, it should give you an idea of how competitive the Turion 64 will be performance-wise, with the Pentium M.

As you can see, AMD should have no problem remaining performance competitive with the Pentium M, but there are obviously many other factors that aren't depicted in the article mentioned above. Mainly we have no idea how the Turion 64 will fare in a power consumption comparison, or how competitive it is from a form factor standpoint. AMD has been aiming at the thin-and-light market from the start with Turion 64, but there's no guarantee that the Turion 64 can get into as thin and as light notebooks as Centrino. AMD has always been one step behind Intel when it came to chip packaging, which carries a lot of weight in form factor discussions.


Are you saying Turion against Dothan or Turion against Yohna (as Yohna is Core Duo and Core - single core Yohna)???[/quote]


Like I said, this is a hobby, not a job. I could be wrong as to which site did the review, but I was surprised becuase the Dothan beat desktop chips. I'll look for the review. The point still remains that mobile would be the best move for 65nm right now.
April 2, 2006 6:45:07 AM

Quote:


Above 500GB, the fastest systems are 64 proc machines. The 4 WAY SPACE IS OWNED BY AMD.

The only benchmark using single 4S systems are at 100GB for TPC-H. The higher MB tests all use clusters. The HP 4S is faster than the Dell system, but Dell uses the slower Intel Truland platform.

And seeing as there's only been 8 new TPC-H submission in the last year, vs 47 for TPC-C, it's quite obvious which one is more important.

Quote:

The only reason Dell wins in TPC-C is because they get the best deals on Intel HW.

Its IBM who has the fastest 4S x86 solution, because of their chipset which outscales glueless Opteron. And you can get it up to a 32S.

Single-core 4S:
IBM x 366 with 3.66GHz Xeons: 150,704
HP DL585 with 2.8GHz Opteron: 138,845

Dual-core 4S:
IBM x 460 with 3GHz Xeon: 273,520
HP DL585 with 2.4GHz Opteron: 236,054

IBM also has results for single-core 8S and dual-core 16S systems. Unisys also has 8S and 16S results for their Xeon MP systems

And what does Dell's cost have to do with performance?

Quote:

The fact that it takes CLUSTERS AT 64P to beat it shows who's running DB OLAPOLTP

What are you talking about? The 64 Xeon processor IBM x346 cluster beats the 48 Opteron processor HP DL585 cluster by 50%.
April 2, 2006 6:53:56 AM

Quote:
The fact that it takes CLUSTERS AT 64P to beat it shows who's running DB OLAP/OLTP

What are you talking about? The 64 Xeon processor IBM x346 cluster beats the 48 Opteron processor HP DL585 cluster by 50%.




Isn't that what I said. In 4P or 8P clusters, AMD wins hands down. They will catch up when they intro L3 for 8P+. I guess you're one of those people who has to be right all the time.
April 2, 2006 6:56:45 AM

Quote:


Above 500GB, the fastest systems are 64 proc machines. The 4 WAY SPACE IS OWNED BY AMD.

The only benchmark using single 4S systems are at 100GB for TPC-H. The higher MB tests all use clusters. The HP 4S is faster than the Dell system, but Dell uses the slower Intel Truland platform.

And seeing as there's only been 8 new TPC-H submission in the last year, vs 47 for TPC-C, it's quite obvious which one is more important.

Quote:

The only reason Dell wins in TPC-C is because they get the best deals on Intel HW.

Its IBM who has the fastest 4S x86 solution, because of their chipset which outscales glueless Opteron. And you can get it up to a 32S.

Single-core 4S:
IBM x 366 with 3.66GHz Xeons: 150,704
HP DL585 with 2.8GHz Opteron: 138,845

Dual-core 4S:
IBM x 460 with 3GHz Xeon: 273,520
HP DL585 with 2.4GHz Opteron: 236,054

IBM also has results for single-core 8S and dual-core 16S systems. Unisys also has 8S and 16S results for their Xeon MP systems

And what does Dell's cost have to do with performance?

Quote:

The fact that it takes CLUSTERS AT 64P to beat it shows who's running DB OLAPOLTP

What are you talking about? The 64 Xeon processor IBM x346 cluster beats the 48 Opteron processor HP DL585 cluster by 50%.




BTW, IBM wins TPC-C with the Power5 not the Xeon.
April 2, 2006 7:15:00 AM

Quote:

Isn't that what I said. In 4P or 8P clusters, AMD wins hands down. They will catch up when they intro L3 for 8P+. I guess you're one of those people who has to be right all the time.

That probably has to do with most companies not really caring anymore about the small benchmarks in particular, and TPC-H in general. Look at the age of the submissions.

Quote:
BTW, IBM wins TPC-C with the Power5 not the Xeon.

That's true, the Power5 servers are the fastest outright for TPC-C. That has no bearing on the fact that their X series servers using Xeon MPs and their X3 chipset are the fastest x86 servers.
!