Dual Core 65nm Preview at Tom's HG

[phaxmohdem]

Just pointing out an article at THG that benchmarks a new 65nm Presler Dual core (900 series). Nothing really mind blowing here.

One thing I notices is that 2MB cache per core = does jack $hit for performance in general. Even the conclusion states, that while the new dual core wil put out less heat and allow higher clocks, it does not address the performance leadership of the X2 AMD fanbois smile now.

http://www.tomshardware.com/cpu/20051010/index.html

 

[Cooler]

Well looks like intel has done nothin to help its perfromence in last 6 months.
The whole Presler /Cedar Mill has done almost nothing but a die shrink. I pity anyone who thinks this is much better then smithfiled or 600s.

 

[ahock]

Yeah no improvement at all.... I think the businesses will benefit from this knowing that it has VT already.... All enthusiast can look for is the overclocking potentials..... Do you have any chipsets features?

 

[Lithan]

When is the mobile dual core due out again?

 

[Cooler]

End of this year Yohna looks like best chip Intel has for next six - eight months.

 

[Corporate Thug]

intel is making baby jesus cry

 

[jazzboy]

Well its seems like that extra cache only really made a difference in about 1 app. I wonder if Intel decided to increase the latency like they did with 5xx -> 6xx cpus.

However I will give them credit for lowering power consumption by a good noticable amount. But they should still be much more aggressive with their speedstep - whats so good about having all of their processors idle at 2.8, why not something much lower like 2ghz or 1ghz?

 

[Lonyo]

People say no improvement, but there is, and you know exactly what the improvement is.
Lower power consumption and heat output.
It's pretty much like AMD's move to the Winchester core for AMD64, or from the Palomino to the T-Bred. I'm not totally sure what people were expecting.

 

[Duvie]

It did have modest gain in winrar and that is likely to adopting the different core structure in terms of cache communications...

The real story for the presler is the reduction in power...Intel did not do anything revoltuionary but buy themselves some more headrrom to clock higher as they get a real answer for the X2s and AMD.....Albeit the reduction is not as much as I had expected. We will have to see how much higher it actuially does go and if they really got a handle on the leakage issues of the prescots.

 

[Powermoloch]

Originally posted by: Corporate Thug
intel is making baby jesus cry

I thinks that saying is overused ?

 

[carlosd]

kill Netburst!!, intel please!!!

 

[MBrown]

Was that a THG article?

"What the new chip cannot do, however, is pose a real threat right now to AMD's Athlon 64 X2 family in the area of performance leadership. The Pentium D 900 certainly is the much better product when compared to the 800 series, but it mainly introduces a number of improvements that were overdue."

 

[stevty2889]

The only thing I find good as that they will work with the 945/955 chipset motherboards, so at least the few current Pentium-D users can upgrade without switching motherboards. You can get reasonably priced 845 chipset boards, so with any luck my next loaner chip will be a presler, just so I can try it out.

 

[Cooler]

Originally posted by: MBrown
Was that a THG article?

"What the new chip cannot do, however, is pose a real threat right now to AMD's Athlon 64 X2 family in the area of performance leadership. The Pentium D 900 certainly is the much better product when compared to the 800 series, but it mainly introduces a number of improvements that were overdue."

Wait so Slaping on 2MB more 65 nm shirnk and 200 Mhz makes this so much better the 800s? It smells like another intel rush job to me.

 

[Hacp]

Originally posted by: Cooler

Originally posted by: MBrown
Was that a THG article?

"What the new chip cannot do, however, is pose a real threat right now to AMD's Athlon 64 X2 family in the area of performance leadership. The Pentium D 900 certainly is the much better product when compared to the 800 series, but it mainly introduces a number of improvements that were overdue."

Wait so Slaping on 2MB more 65 nm shirnk and 200 Mhz makes this so much better the 800s? It smells like another intel rush job to me.

Obvous AMD fanboy. Reduced heat means higher clocks, which meants more ocing headroom. If people can get 4GHZ on air, that means that it will be competative with the 2.5GHZ-2.6 GHZ x2s that are floating around. Heres to hoping thers a new 2.8 GHZ model of the core .

 

[Duvie]

I always though cache was expenisve and added to heat...What a waste if this is the type of gains it got ....256 to 512 was big and even 512 to 1mb was nice...however 1mb to 2mb was anemic and the 2mb to 4mb seem like more unused potential...

I imagine there are some apps like superpi and other that may like this...the question is is this the same 6xx series cache??? You know the cache that was slower then the 5xx series cache with its increased latency...

 

[stevty2889]

Most likely high has the same slower cache of the 6xx series. Cedar Mill is just a 65nm 6xx series after all.

 

[Cooler]

Originally posted by: Hacp

Originally posted by: Cooler

Originally posted by: MBrown
Was that a THG article?

"What the new chip cannot do, however, is pose a real threat right now to AMD's Athlon 64 X2 family in the area of performance leadership. The Pentium D 900 certainly is the much better product when compared to the 800 series, but it mainly introduces a number of improvements that were overdue."

Wait so Slaping on 2MB more 65 nm shirnk and 200 Mhz makes this so much better the 800s? It smells like another intel rush job to me.

Obvous AMD fanboy. Reduced heat means higher clocks, which meants more ocing headroom. If people can get 4GHZ on air, that means that it will be competative with the 2.5GHZ-2.6 GHZ x2s that are floating around. Heres to hoping thers a new 2.8 GHZ model of the core .

Im no ones fan boy I just call it as I see it. My current Rig is Intel if you did not notice. All I know is Intel should have made the L2 On cache be able to commucate with each other not from over the fsb . You would think after simthfiled they would have learn that is not the way to go.

 

[Hyperlite]

Originally posted by: stevty2889
The only thing I find good as that they will work with the 945/955 chipset motherboards, so at least the few current Pentium-D users can upgrade without switching motherboards. You can get reasonably priced 845 chipset boards, so with any luck my next loaner chip will be a presler, just so I can try it out.

THAT is a miracle.

 

[Corporate Thug]

Originally posted by: Powermoloch

Originally posted by: Corporate Thug
intel is making baby jesus cry

I thinks that saying is overused ?

i agree, thats why i use it so often. i love beating $hit to death

 

[Viditor]

Originally posted by: Duvie
I always though cache was expenisve and added to heat...What a waste if this is the type of gains it got ....256 to 512 was big and even 512 to 1mb was nice...however 1mb to 2mb was anemic and the 2mb to 4mb seem like more unused potential...

I imagine there are some apps like superpi and other that may like this...the question is is this the same 6xx series cache??? You know the cache that was slower then the 5xx series cache with its increased latency...

Cache is VERY expensive (it adds more to the die size than the core by about 2 to 1), but doesn't necessarily add more heat...by adding cache, you may increase the aggregate heat, but you decrease the number of localized "hot spots" on the die.

One thing to note, the new Intel dual cores (900 series) will almost certainly be more expensive than their 800 series brethren...

 

[DrMrLordX]

Originally posted by: Viditor

One thing to note, the new Intel dual cores (900 series) will almost certainly be more expensive than their 800 series brethren...

That's a bit of a problem. The Pentium D 8xx chips only have one main advantage, and that's their low price. If you gave me a P-D with reduced thermals and similar or slightly higher clock speeds, I would actually consider one as such a CPU might amount to a sleeper budget OC champion. Let's face it, the 820D is tempting, but the price difference between it and the X2 3800+ will probably be consumed by the bizarre cooling setup you'd need to challenge the X2's performance. You'd need to hit maybe 4-4.5 ghz on a 820D to really give an OCed X2 a run for its money.

Strapping on extra cache as a marketting gimmick will probably keep the price of the 9xx series too high for it to be competative as a buget OC chip.

 

[stevty2889]

I don't think they will cost much more. The cache is larger, but the die size is smaller due to the 65nm proccess, so it doesn't really cost more to produce the chips, other than the cost of the new proccess.

 

[Ready]

126 watts idle? No thanks

 

[Viditor]

Originally posted by: stevty2889
I don't think they will cost much more. The cache is larger, but the die size is smaller due to the 65nm proccess, so it doesn't really cost more to produce the chips, other than the cost of the new proccess.

True, but remember that 1MB of cache is about 40-50% larger than the core...these will have a total of 4MB cache.

Edit: I suppose that if the 65nm shrink does cut the core in half (as well as the cache transistors), the Presler with 4MB will be about the same 209mm2 as the Smithfield...but for a new process that will be very expensive! Remember that any new process takes time to ramp, and with a large die it takes even longer to get good yields (so many chances for a problem on the die). This means that with Presler will probably be a very expensive CPU to manufacture for the first 6 months - 1 year.