Core i3 vs. Phenom II X3 "clock-4-clock"

[Nedjo]

http://en.inpai.com.cn/doc/enshowcont.asp?id=7488

[informal]

Thanks for sharing the link.
The L3 on Phenom II X3 is a big help as we can see from the charts. Also physical cores seem to scale pretty good,no more odd results with 3 cores CPUs in desktop workloads - as we can see the X3 is almost always on top Vs the same clocked i3. There are some confirmations that there will be another BE version of Phenom II X3,a 740 running at 3Ghz .Along with 800 series chipset ,I can't see how i3 with its IGP can compete with it.

[Manicdan]

i think the funny part is how thats a 1 year old first edition PII.

[ajaidev]

In terms of IGP any Intel IGP "even 900mhz one" can be defeated by a 4200 with side-port so this combo could take on a i3 530 very successfully

AMD Phenom II X3 720 Black Edition + ASRock A785GM-LE/128M

On one hand you have $110 + $65 = $175 on another $113 + $85 = $198

[Hornet331]

i think the funny part is how thats a 1 year old first edition PII.

And how exactly is that funny... more cores > less cores in MT apps. (as always)

Its quite interesting to see that 2+2SMT threads can keep upo with 3 full cores... would be nice if there would also have been results without HT.

[Nedjo]

with Athlon II X4 630 on the market, and with price lower than Ph II X3 720, I don't see the need for 740BE...

it's a pity that X4 630 wasn't included in that review. especially 'cos of its default frequency of 2.8 GHz.

[informal]

And how exactly is that funny... more cores > less cores in MT apps. (as always)

Its quite interesting to see that 2+2SMT threads can keep upo with 3 full cores... would be nice if there would also have been results without HT.

Don't forget those "2+2SMT" threads are actually based on 32nm 2nd gen of Nehalem,while here we have a 45nm X3 Deneb based core as a competitor. The cores themselves(if done on the same node) are taking up considerably different die areas,24.4mm2 for Nehalem and 15.3mm2 for Deneb meaning X3 actually has less total logic die area compared to Nehalem(plus Westmere cores if done on 45nm due to some improvements would end up probably a bit bigger!).X3 Deneb is doing great against the Westmere SMT dual cores.

[kiwi]

The problem is that x3 wont do 4.5ghz stable on air, not even close

[Hornet331]

Don't forget those "2+2SMT" threads are actually based on 32nm 2nd gen of Nehalem,while here we have a 45nm X3 Deneb based core as a competitor. The cores themselves(if done on the same node) are taking up considerably different die areas,24.4mm2 for Nehalem and 15.3mm2 for Deneb meaning X3 actually has less total logic die area compared to Nehalem(plus Westmere cores if done on 45nm due to some improvements would end up probably a bit bigger!).X3 Deneb is doing great against the Westmere SMT dual cores.

All true but in the end only total die space counts, and since heka is a salvaged quadcore your looking at 258mm² vs 81mm² (+ 114mm² gpu)

[zalbard]

Yes, I think the performance of the overclocked chips would be interesting to have a look at.

[Macadamia]

All true but in the end only total die space counts, and since heka is a salvaged quadcore your looking at 258mm² vs 81mm² (+ 114mm² gpu)

That's the reason why AMD's switching to Propos even for X4s.

~ ASP, 100 mm^2 less.

160mm^2 on a steady 45nm process vs 2 dies on 32 with interdie connects needed.. (that one possibly more costly than yields/diearea for the budget market)

[STaRGaZeR]

"clock-4-clock"

More like 3 vs. 2 cores at the same clock speed. Lame, but I guess this is the best comparison you can do considering AMD and Intel offerings nowadays.

[vengance_01]

The I3 and I5's are very nice replacements for Core 2. Just because one has 2 core plus 2 HT and another just has 3 cores, if they are priced the same then the number of core is irrelevant. Both are great budget chips that can handle nearly anything you throw at them. I still think the BE 550 unlocked to 4 cores is by far the best value around at this point.

[qurious63ss]

The thing I see here is that intel wants to make money with i3 while amd could care less if their chips make any profit since these days they don't really care if they make money or not since their sugar daddies will be there to bail them out.

[Manicdan]

i also think AMD should be making the move to L3less chips, the die space is costing them way too much. i could live with <10% less performance for >30% price reduction.

i just think that its funny to compare the first 45nm AMD chips with the first 32nm Intel chips. if they did the comparison right, the intel one should stomp amds in every way. its a year old and on a dying platform. instead we see a near matching comparison. (i am NOT saying they are the same in every way, im just saying the review makes it look that way)

[Nedjo]

"clock-4-clock"

More like 3 vs. 2 cores at the same clock speed. Lame, but I guess this is the best comparison you can do considering AMD and Intel offerings nowadays.

well there's always space for Pentium G9650 vs. Phenom II X2 550 (downclocked @ 2.8GHz) showdown!?

After all these two CPUs cost the same amount of money - 89 EUR:

Pentium G9650 - http://www.arlt.com/hardware+oxid/pc...950+28ghz.html

Phenom II X2 550 Black Edition: http://www.arlt.com/hardware+oxid/pc...ii+x2+550.html


Beyond technical curiosity, in the land of pure value for the money, I don't see how can Pentium G9650 compete with the 66 EUR worth Athlon II X2 250: http://www.arlt.com/hardware+oxid/pc...ii+x2+250.html

Hell that Athlon II X2 250 is the best buy Dual-Core CPU, if you consider the fact that for same amount of money you can only buy outdated Pentium E5300: http://www.arlt.com/hardware+oxid/pc...ii+x2+250.html

[Nedjo]

i just think that its funny to compare the first 45nm AMD chips with the first 32nm Intel chips.

Why it is funny? They've chosen clock parameter. What's funny about that choice?

if they did the comparison right, the intel one should stomp amds in every way.

and the right way would be?

its a year old and on a dying platform.

Sorry dying platform?

instead we see a near matching comparison. (i am NOT saying they are the same in every way, im just saying the review makes it look that way)

well apparently review has shown what you get from 3-core "K10.5" architecture @ 2.8 GHz against HT enabled Clarkdale w/o Turbo @ 2.8, and 2.93 GHz... don't understand what you've expected?

[grimREEFER]

the 2 cpu's seem roughly equal, how do they compare pricewise?

[ajaidev]

i also think AMD should be making the move to L3less chips, the die space is costing them way too much. i could live with <10% less performance for >30% price reduction.

i just think that its funny to compare the first 45nm AMD chips with the first 32nm Intel chips. if they did the comparison right, the intel one should stomp amds in every way. its a year old and on a dying platform. instead we see a near matching comparison. (i am NOT saying they are the same in every way, im just saying the review makes it look that way)

If the AMD ones were x2's and not x3 it would have been stomped.

Also talking about athlon X4's the 620 does work out to be very comparative with the i5 661 than the i3 530 like in these benchs:

the 2 cpu's seem roughly equal, how do they compare pricewise?

i3 530 - $113
X3 720 - $125
X4 630 - $115
X4 620 - $99

Cheapest 785 with side port - $65
Cheapest P55 - $85

[xlink]

can we compare at 4.2Ghz on the i3 and 3.6GHz on the PII x3?

[Dr_Swizz]

i also think AMD should be making the move to L3less chips, the die space is costing them way too much. i could live with <10% less performance for >30% price reduction.

AMD is already making such chips. The dual core Athlon IIs and the Sempron 140 have 1MB L2 and no L3 cache.

[Manicdan]

Why it is funny? They've chosen clock parameter. What's funny about that choice?

and the right way would be?

Sorry dying platform?

well apparently review has shown what you get from 3-core "K10.5" architecture @ 2.8 GHz against HT enabled Clarkdale w/o Turbo @ 2.8, and 2.93 GHz... don't understand what you've expected?

so much hate in your replies.

the right way is the way in which they would do it if their goal was to show how much better it could be. im not saying this is a wrong way, but its the wrong way if they wanted to make it look better. which a 32nm newer chip should be.

ddr2 is dying, that platform.

i expected to see how 32nm can beat 45nm. profitability, power consumption, pricing, future proofing. the article is lacking any detail needed to show why one person would pick one cpu over another. the article isnt AMD biased, otherwise it should have used something they are still making (i do believe they stopped making the 940 and 720 mid last year?) so why not compare to a 435?

[Mats]

Manicdan: They used DDR3 in both platforms, not DDR2.

Asking questions isn't the same thing as hate.

[Nedjo]

Asking questions isn't the same thing as hate.

thanks for pointing that out Mats!

[FlawleZ]

And how exactly is that funny... more cores > less cores in MT apps. (as always)

Its quite interesting to see that 2+2SMT threads can keep upo with 3 full cores... would be nice if there would also have been results without HT.

The only fair way to compare clock for clock, core for core is to disable HT on the i3 and compare to 550BE @ 2.8Ghz. Then again, the only relevance here is price. If the BMW 335i was the same price as Honda's Accord, you might say that's an unfair comparison. I would call it checkmate.

Of course physical cores>logical cores. But the HT has long been Intel's bread and butter and if you're going to market its multitasking prowess then you better have the results to back it up.