Anandtech i5 preview

[Chrono Detector]

Pretty nice, performs quite close to the Core i7 920. Hope the price isn't too bad, definitely interested in this one.

[orangekiwii]

lga1366 is the ENTHUSIAST segment and will remain that way for several years up until AT LEAST sandy bridge...

core i5 is designed to be the lga775 for a while...

lga1366 remains the top performer and choice for enthusiests with its added features and what not but

lga1156 is the choice for mid end even high-mid end builds



thats how I see it... Anything else is too stupid to even make sense

[Sly Fox]

Intel has said from day 1 that i7 was the high end and i5 was mainstream...

Guess some people ITT have some reading to do. Maybe some opinions to reevaluate.

[rk7p5]

So what's the price going to be like for i5? Say you're ugrading and just need mobo, ram, and cpu, how much would be saved vs i7?

[Glow9]

Intel has said from day 1 that i7 was the high end and i5 was mainstream...

Guess some people ITT have some reading to do. Maybe some opinions to reevaluate.

Yet some 775's can beat i7s.... This is getting kinda dumb with some people saying i5's can beat i7's. Whats the point of having something in between 775's and i7's when 775's can again beat some i7's. Too me this just seems stupid it's a cash grab as far as I'm concerned.

I'm guessing an i5 will cost less then high end 775's. Which means people will opt out of going for $$ 775's and will go for i5's which also means they will buy P55's. If you ask me this is just a slap in the face to the consumer. I rather have P55's w/DDR3 & 775's.

[CedricFP]

So what's the price going to be like for i5? Say you're ugrading and just need mobo, ram, and cpu, how much would be saved vs i7?

Apparently, if P55 mobos are similar in pricing to the lower-end X58 mobos, not much.

[grimREEFER]

Yet some 775's can beat i7s.... This is getting kinda dumb with some people saying i5's can beat i7's. Whats the point of having something in between 775's and i7's when 775's can again beat some i7's. Too me this just seems stupid it's a cash grab as far as I'm concerned.

I'm guessing an i5 will cost less then high end 775's. Which means people will opt out of going for $$ 775's and will go for i5's which also means they will buy P55's. If you ask me this is just a slap in the face to the consumer. I rather have P55's w/DDR3 & 775's.

the qx9770, the fastest 775, is slower than the core i7 920, the slowest of the core i7.

[kl0012]

Apparently, if P55 mobos are similar in pricing to the lower-end X58 mobos, not much.

Taking in account that P55 = P45 = $40 and P55 based mobos will be cheaper in production (because of less components and simplified layout) I expect very reasonable price for mid/low end boards (~$100-120). May be not immediatly after release, but at the end of the year I hope to see some P55 boards even cheaper then low end P45.

[Glow9]

the qx9770, the fastest 775, is slower than the core i7 920, the slowest of the core i7.

I see tests where the i7 is far from dominated

[naokaji]

Yet some 775's can beat i7s.... This is getting kinda dumb with some people saying i5's can beat i7's. Whats the point of having something in between 775's and i7's when 775's can again beat some i7's. Too me this just seems stupid it's a cash grab as far as I'm concerned.

Yes, C2D / C2Q does match or beat I7 in some rare scenarios, but that is when using ancient software only.
I5 will beat I7 in some situations, simply because it will end up with higher clocks when not all cores are fully loaded due to turbo giving more than +1 multi.

Just because desktop software developers have not caught up yet with the fact that netburst and single cores are dead does not mean intel is not delivering enough improvement, look at some benches wth server software, nehalem wipes the floor with everything else, so really, it is just a question of time until I5 and I7 will be able to show the real power.

[saaya]

By the time this comes out, it will be close enough to 12 months since i7 was launched and we get F#$% all improvement.

ive been warning people even before 1366 launch that they shouldnt wait for 1156 if they want perf...

He's comparing the 2.13 GHz i5 to the 3.2 GHz X4 955. The i5 is better performing in 8 benchmarks, and the X4 is the faster one in 6 benchmarks.
Dunno if those benchmarks are well chosen, someone else have to comment on that.

but the 2.13ghz chip was oced to 2.66, thats quite important to mention :P

AMD needs to compete with i7 first

huh? why?
no they dont, and they dont even try or plan to... i7 is a niche, theres not that much money to be made there, amd wants large volume, mainstream and entry level... so its clearly i5 vs phenom2

Apparently, if P55 mobos are similar in pricing to the lower-end X58 mobos, not much.

yepp, its possible to create x58 boards that cost around 100$ as well... basic entry level, same as p55 basic entry level for 100$...

[geo]

Its not even a 2.66Ghz but an Oced 2.13Ghz...also its an engineering sample. Turbomode doesnt work correctly etc.

Thanks mate overlooked that part

[Chad Boga]

amd wants large volume, mainstream and entry level... so its clearly i5 vs phenom2

Have AMD confirmed they will be bringing out higher clocked PhII's on 32nm that will run in existing AM3 boards?

[informal]

Have AMD confirmed they will be bringing out higher clocked PhII's on 32nm that will run in existing AM3 boards?

They have not confirmed anything,but you can bet they will not make Bulldozer as their first 32nm MPU.32nm is perfect for 6/8 core tryout on K10 before they go with the more complex and new design.

[Chad Boga]

They have not confirmed anything,but you can bet they will not make Bulldozer as their first 32nm MPU.32nm is perfect for 6/8 core tryout on K10 before they go with the more complex and new design.

A 32nm Quad PhII at 4.0Ghz stock standard retail release would be nice, very nice.

Dunno if it will eventuate, and I hope they don't just use their first 6 months of 32nm for Hex Core.

[Mats]

but the 2.13ghz chip was oced to 2.66, thats quite important to mention :P

Anandtech uses three configurations for the i5: 2.13 GHz HT, 2.66 HT, and 2.66 GHz no HT.

The i5 2.13 GHz is faster than the X4 3.2 GHz in 8 of 14 benchmarks.
LOOK AGAIN.

[informal]

A 32nm Quad PhII at 4.0Ghz stock standard retail release would be nice, very nice.

Dunno if it will eventuate, and I hope they don't just use their first 6 months of 32nm for Hex Core.

Yeah QC with high(er) clocks would be great,but 4Ghz is a bit too optimistic IMO,but not impossible of course.
Somehow I think they will do just what you hope they don't,use it for hex core .Server sales will need it badly and the chip will be rather small compared to Istanbul.

[Chad Boga]

Yeah QC with high(er) clocks would be great,but 4Ghz is a bit too optimistic IMO,but not impossible of course.

I think they will get 3.4Ghz out of 45nm, with 3.6Ghz an outside possibility(but I don't expect to see it), so eventually releasing a 4.0Ghz on 32nm doesn't seem that unrealistic.

Somehow I think they will do just what you hope they don't,use it for hex core .Server sales will need it badly and the chip will be rather small compared to Istanbul.

Why can't they do both? Even if they spend the first 3 months on just Istanbul alone.

[mAJORD]

I personally think 3.4 before the end of the year (around launch of i7 I bet) , mayyybe 3.6 before 32nm IF they spin a new stepping before then (under the pressure of i5 especially).

I also think 6 core on the desktop is a possiblity for 32nm. They can use speed bumps to compete with non HT i5's (going by anand's review) but with HT on, contrary to what everyone insists performance is much higher in the types of benchmarks used by the likes of Anand.

The next person to tell me HT isn't the reason i7 owns in reviews can stfu.. Anand just put it to you in plain figures.. 18% faste averaged across all tests.

-edit, applogies, not all tests it seems. But i maintain HT is a big advantage
Don't confuse that with a reason i7 beats PHenom II.. clock for clock thread for thread its still some 20% faster

[kl0012]

-edit, applogies, not all tests it seems. But i maintain HT is a big advantage
Don't confuse that with a reason i7 beats PHenom II.. clock for clock thread for thread its still some 20% faster

+1
A simple calculation (based on Anand's tests) shows average 9% performance advantage of i5-2.66GHz-No-HT over PII-3.2GHz

[RPGWiZaRD]

+1
A simple calculation (based on Anand's tests) shows average 9% performance advantage of i5-2.66GHz-No-HT over PII-3.2GHz

Exactly, i5 won't even compete performance wise with AMD Phenom II like many presumed but I weren't one of them though and expected it to be just behind i7. I'm just confused why Intel even launched i5 tho and didn't offer some cheaper chipset than X58 for i7, for example P55. That had made more sense IMO. Well maybe if there's not gonna be any decent upgrade options for i5 so they get additional cash from people having to switch socket again but then again that would be tactically stupid move to upset so many customers. I cannot imagine they'd do such thing, that they haven't revealed its future plans for socket 1156 is probably to try make a better balance between i7 and i5 sales, revealing some planned future upgrades now for i5 would make i7 sales rockbottom when Lynnfield is launched. I'm pretty certain there will be some but the plans aren't revealed this year.

[Shintai]

Exactly, i5 won't even compete performance wise with AMD Phenom II like many presumed but I weren't one of them though and expected it to be just behind i7. I'm just confused why Intel even launched i5 tho and didn't offer some cheaper chipset than X58 for i7, for example P55. That had made more sense IMO. Well maybe if there's not gonna be any decent upgrade options for i5 so they get additional cash from people having to switch socket again but then again that would be tactically stupid move to upset so many customers. I cannot imagine they'd do such thing, that they haven't revealed its future plans for socket 1156 is probably to try make a better balance between i7 and i5 sales, revealing some planned future upgrades now for i5 would make i7 sales rockbottom when Lynnfield is launched. I'm pretty certain there will be some but the plans aren't revealed this year.

They already revealed 32nm for both sockets. Not sure how far into the future you want? LGA1366 and LGA1156 follows hand in hand. LGA1156 will outlive LGA1366 aswell.

[Smartidiot89]

There is c0 - d0 and the new cpu's that come with the d0 rev, that's enough.

AMD needs to compete with i7 first, 'i5' fills in for c2q so they have their upper, mid and extreme end desktop tier covered. With amd announcing no more new cpu's for a while intel can just work towards their 'tock'. AMD might have to tweak their pricing again to bring the field level.

AMD won't have a chance in the high-end segment, the company has said so themselves. Until Bulldozer arrives AMD will be forced to rely on price/performance.

[kl0012]

Exactly, i5 won't even compete performance wise with AMD Phenom II like many presumed but I weren't one of them though and expected it to be just behind i7. I'm just confused why Intel even launched i5 tho and didn't offer some cheaper chipset than X58 for i7, for example P55. That had made more sense IMO. Well maybe if there's not gonna be any decent upgrade options for i5 so they get additional cash from people having to switch socket again but then again that would be tactically stupid move to upset so many customers. I cannot imagine they'd do such thing, that they haven't revealed its future plans for socket 1156 is probably to try make a better balance between i7 and i5 sales, revealing some planned future upgrades now for i5 would make i7 sales rockbottom when Lynnfield is launched. I'm pretty certain there will be some but the plans aren't revealed this year.

As usual it was driven by marketing reasons. But since S1366 targets high end (which was known for a long time) I didn't bother even think about buying I7 . I think Intel defined pretty clear its market segments and it may be a good thing.

[JumpingJack]

great performance...

so Anandtech isn't under NDA?

He is only under NDA when he obtains his samples directly from Intel and signs the NDA for that product.

He did not sign an NDA for this sample, rather the person who supplied him the sample broke their NDA, hence the reason Anand blacks out the processor serial numbers and hides the capacitors.

On a side note -- the most interesting about this 'preview' isn't the performance, but the fact that he could OC with Bclk. Meaning the PCIe clock on the chip is independent. There was some rumor grumbling (Fuad of FUDzilla) that OC would be limited because the proc would clock to the PCIe clock. Which, of course, is now shown to be nothing but rubbish.