Core i3/i5 review @ anandtech

[grimpr]

What? Yeah they were surprised but more by the bad performance/price ratio of the 661 - don't even think about the 670.

And I'm not sure if you've read the review, but "The Core i5 661 is priced entirely too high for what it is" or my personal favorite "I actually had to call Intel and ask them to explain why this wasn't a worthless product", do not sound THAT impressive.
And I'd be interested if you told us, how exactly "drivers" should increase the performance of a CPU.. not that you ever would or could.


The i3s are interesting, but the new i5s are just completly overpriced. Yeah you can overclock them a big deal (at least if you have a year supply of Ln2), but that doesn't help them a lot if you look around the reviews @6ghz.

Its a pointless talk since you cant change his mind at all. In his world, the worthless competitor to Intel and its products are the main obstacle for buying the SKU he wants, the F.T.C. is secretly manipulated from green hooded european masons at AMD, Anand has proven unworthy and suspicious despite borrowing his last sentence and screaming it as truth "Its the FTC because of AMD you stupids". A global conspiracy after all, the failed socialist E.U., then the F.T.C., next is Putins Russia pointing its nukes at Santa Clara.

Neglecting that Intels cpu product dept. are not morons by cannibalizing the sales of 3-4 entire cpu market segments by offering the object of his affection,the i5 670, for 100$. He just wishes to have been this way just to satisfy his burning hatred for all non-Intel.

Good Night, and Good Luck.

 

[Nemesis 1]

Who are you talking about? How could Virtuallarry understand so well and you missed it so completely?

 

[lopri]

I don't have any opinion on the conspiracy theory, but wasn't the pricing of these chips known (semi-officially) way before the settlement or the new suit? I thought I saw the forecast prices last summer.

 

[Nemesis 1]

Yes you did Lopri . But what we know publicly now has been said long ago , Its in the law suite I believe. Intel did come to agreement on more than just $$$. i5 pricing as you see it is the result of such an agreement . You are aware of what the EU said . About intel holding back innovation and hurting consumers . All any of us has to do is take a stroll down memory lane . What were CPU pricies in 1990? what were they in 1999? than again in 2005??? Than what were they after C2D was released . Someone is lieing here. It does not appear to be Intel by all accounts. Youve been around a while you know this is true . As for innovation were is it . Seems to be on Intels side of the argument . C2D proved the FSB was very good up to this last generation ,does it not?

Look at the ATI nividia arguments from the 600 till the present almost all said Ninidia had the better tech . But what just occurred ATI proved that not to be the case.

Now with the FTC in here I can't wait to see how they muck this up further.

 

[piesquared]

Take a wild wild guess what will happen in the real world instead of your "ifs" and from those "sheeps" that you called.

Nobody cares for the vapourware and expensive Hydra chip, in my analysis a destined failure attempt backed by Intel Capital money to get rid for those sli/crossfire licence fees it pays and like the the "one ring" bind the competitors gpus on its platform, noone serious enough will drive multiple GPU's for gaming/folding/gpgpu with a little dualie and almost all people driven by the natural "more cores are the better deal" will snag the Quads in a snap. Make it also that multithreaded software is continually doubling its numbers in apps and scaling in games and you have your breakfast served. Enthusiasts dont buy hype anymore, they want perfect execution on all fronts and they knew that Clarkdale is not gonna cut it until Sandy Bridge arrives. By then, it will be Quadcore from the bottom up and Hyperthreading will be for free on all SB skus and not priced as a premium feature to differentiate the lines. We wont bother to talk about AMD's future products since many are pathological allergic to the name,brand and its products.

Wise and experienced enthusiasts know already their next upgrade, from the i5 750 or the 920 to a 6/8 core Sandy. Full Tock. AVX instructions, 256bit engines/cylinders, mature 32nm. rock n roll.

Just thought i'd suggest that AMD's faster maturing processs needs to be considered also.

 

[nyker96]

It's sad to see that above 200 mark Intel is just competing against itself, sucks to be the lonely top dog. i5s i7s 600s 700s 800s 900s all just competing against each there. amd is been compressed down to the very bottom. I think if Intel wanted, it can easily kill off Amd cpu business right now. can IBM or someone else step in and provide more competition here? Intel is like 1,5 years ahead of amd in process and core technology. i think only way for amd to have a chance is if intel do another P4, well that don't happen often, even if they did Intel got enough cash reserve to not feel too much sting from such a mistake.

 

[Nemesis 1]

You should have also asked about ATI cross fire license fees that do not exist from a learned enthusiasts. Ya I to want to see the BD go up against SB. But I am still on a P4c for browsing . I can build a new i670 or 661 depending on how the Vapor ware hydra chip performs for 1/3 the cost of the orginal build of this browser. Even tho this does everthing I need it for. As for video editing and gaming this thing was done along time ago in the arenea. I like what i5 670 is offering . It will be an ass kicking browser. Am I not allowed to make that choice or anyone else and still have fun o/c and running benchies on a browser . I like that . The power requirments of both my wifes machine and my gamer just are not green for browsing the net . Complete waste in fact.

 

[piesquared]

I think this is just the beginning of the marketing disastors intel is going to face with it's much vaunted tick tock strategy.

 

[piesquared]

It's sad to see that above 200 mark Intel is just competing against itself, sucks to be the lonely top dog. i5s i7s 600s 700s 800s 900s all just competing against each there. amd is been compressed down to the very bottom. I think if Intel wanted, it can easily kill off Amd cpu business right now. can IBM or someone else step in and provide more competition here? Intel is like 1,5 years ahead of amd in process and core technology. i think only way for amd to have a chance is if intel do another P4, well that don't happen often, even if they did Intel got enough cash reserve to not feel too much sting from such a mistake.

I think you underestimate Bulldozer, Bobcat and AMD's monster line of Fusion chips. Not to mention ATi's next creation.

 

[Nemesis 1]

Got any benchmarks

 

[s44]

what do you mean they won't realse quads on 32nm? I'm faily sure they will probably end of the year for both 1156/1366 platforms.

Unless Intel's revised the roadmap really recently, 1156 is going directly from 45nm quads to EOL (replaced by Sandy Bridge). The current batch of dual-cores are the only 32nm parts the socket is going to see.

Monopoly power: got to love it. (Speaking of which, this forum does have an ignore list. It's very useful for certain threadjacking posters.)

 

[DrMrLordX]

http://www.bit-tech.net/hardware/cpus/2010/01/04/intel-core-i5-661-core-i3-530-cpu-review/9

Here is a comparison of 32nm dual core overclocked compared to Phenom II x4 overclocked in Crysis.

Very interesting . . . the entire article has some good info.

The downside is that the bit-tech reviewers couldn't get their i3-530 beyond 3.5 ghz. However, it certainly did put on quite a show against the X4-630, which I must admit was quite surprising, especially in the H.264 benchmark. And all that against a 630 at 3.7 ghz (which is an unexpectedly high overclock for Propus).

Then the X4 965BE came along and crushed them both, which again surprised me . . . the L3 made a huge difference for Propus/Deneb.

The i3-530 also didn't seem to have much trouble with scaling up to 3.5 ghz, but at that speed, I suspect the BCLK/memory speed limitation Anandtech reported had not yet been reached. Also, look at the QPI multi that bit-tech used: x32! Anandtech never bothered using anything beyond x24.

I still think the 630 could have turned in a better showing with better memory (bit-tech used DDR3-1600 with unspecified timings), but then I am also of the opinion that Propus chips need a system memory latency of 50-70 cycles which may not be possible with current DDR3 unless your CPU core speed happens to be rather low.

I would like to see what sort of results Anand gets with his i3 once he gets around to overclocking one. Hopefully he will provide results at stock and overclocked speeds so we can get more data points with which to judge its capacity to scale properly.

Nevertheless, it seems that i3's cache architecture serves it well enough, despite the funky "integrated" memory controller.

Theoretically there is no way a physically closer memory controller being higher latency than motherboard memory controller.

Yes and no . . . Anand's review seems to indicate that the IMC/GPU element of the package utilizes a memory controller derived from the p45 chipset. He doesn't seem to go into much detail about this and instead vaguely indicates that it just isn't designed to work with QPI. It makes me wonder if the memory controller requires some sort of hardware bridge to work with QPI - in other words, something that would probably impose a built-in performance penalty, which is why the Clarksdale featured in Anand's article turns in inferior memory latency to that of a Core 2 Duo, despite having better bandwidth numbers.

 

[Hey Zeus]

Why didn't anyone get a 670 to test? That's all i cared about reading through the benchmarks today.

 

[toyota]

Why didn't anyone get a 670 to test? That's all i cared about reading through the benchmarks today.

why? its just the same thing with a slight speed bump and more expensive price. how can you not get through your head how foolish the 670 is when you can buy the 750 for 100 bucks cheaper??

 

[Voo]

Now we have a topic here on FTC sueing intel. What would the FTC say if intel priced the way YOU THINK they should . and what would that price do to AMD. LOL . Something isn't making any sense here. It isn't me . Tell us were would you price the i5 661 . go ahead say it . than look at what that would do in the only market AMD can compet in . Gee!

I think you should try and read the FTC charges - probably in your language if your writing skills are any indication for your comprehension skills.

The charges against Intel have NOTHING to do with their MSRPs.

The CPU side of the complaints:
1. Intel rewarded OEMs to not use AMD’s processors through various means, such as volume discounts, withholding advertising & R&D money, and threatening OEMs with a low-priority during CPU shortages.
2. Intel reworked their compiler to put AMD CPUs at a disadvantage
3. Intel paid/coerced software and hardware vendors to not support or to limit their support for AMD CPUs.
4. False advertising. This includes hiding the compiler changes from developers, misrepresenting benchmark results (such as BAPCo Sysmark) that changed due to those compiler changes, and general misrepresentation of benchmarks as being “real world” when they are not.

source: http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3690


Oh and you didn't answer my question.. how would "drivers" increase the performance of the i5s? Just shows once more that you've got no idea what you're talking about..

@DrMrLordX: Anand mentioned the problem with the memory.. if they're using a preset table for the clock skew, there aren't that many possibilities to use "better memory". Though that'd probably help the CPU a lot if overclocking.

 

[Nemesis 1]

First and formost you have not read all the FTC charges against intel . Back on topic . Can anyone here explain to this person why updated drivers would help i5 performance . I have no desire to educate him.

 

[Voo]

First and formost you have not read all the FTC charges against intel .

Okay show me a better source, oh and yeah tell me how a driver should increase the performance, I'm really interested in that.

Two very interesting claims, but it's nothing new that you claim something and then never answer it..

 

[Hey Zeus]

why? its just the same thing with a slight speed bump and more expensive price. how can you not get through your head how foolish the 670 is when you can buy the 750 for 100 bucks cheaper??

Could you get it through your head i don't care what you think? It's my money and i'll do whatever i want with it. The chip is an absolute ripoff yet i don't care.

 

[Nemesis 1]

Why does ATI/Nividia update drivers and whats the benefit of those upgrades, Does i5 have have a IGP ondie or not ? As for the FTC I think it best for self I stay ontopic here . Just ask the mods . Actually its best for all that I do not back up such claims as it leads to off topic and in the end after causing much debate I win a cookie. But all ya have to do is topic search .

 

[Phil1977]

Enjoyed reading the reviews!

However there is one test none of the review sites looked into.

I am very curious to see how a 3.33 GHz i5 (with HT AND Turbo turned OFF) compares against the E8600...

I am quite certain that the E8600 is a faster processor (because it has 6mb L2 cache) and that the i5 NEEDS HT and Turbo to actually beat the E8600...

Intel and AMD are using interesting stratgeies. Intel uses a dual core and adds features (Graphics, HT, Turbo) and AMD takes a quad and removes features (No L3 cache).

Now AMD is just about to launch 6 new CPUs. They will have Athlon II x4 2.9 GHz and Athlon II x3 3.0 GHz.

It will be interesting to see which strategy is the better one...

But I know which one is cheaper! 4 cores in 45nm vs. 2 cores in 32 + 1 gpu in 45nm...

Which cpu would I buy? Either the i5 750 or a Phenom II x4. Both are awesome chips, fast at whatever task you throw at them.

What cpu do I have? A cheap Athlon II 250 3 GHz dual core, running at 3.5 GHz

 

[exar333]

Unless Intel's revised the roadmap really recently, 1156 is going directly from 45nm quads to EOL (replaced by Sandy Bridge). The current batch of dual-cores are the only 32nm parts the socket is going to see.

Monopoly power: got to love it. (Speaking of which, this forum does have an ignore list. It's very useful for certain threadjacking posters.)

They already have 32nm quad SKUs for Xeon that will be 1366 compatible (ask Aigo). There will be hexacore 32nm CPUs with 12MB cache, and quad-core 32nm CPUs with the full 12MB cache as well (either 2 cores disabled or faulty). The E5640 looks amazing, and will probably be my next chip this spring. Not sure if 1156 will get these though, 1366 will be king for Xeon compatiblity.

http://www.crn.com/white-box/222001806;jsessionid=U30LVJ3MPH2OJQE1GHRSKH4ATMY32JVN

Exerpt from article: Seven 32nm, quad-core processors are also listed, all with 12MB of cache -- the Xeon X5677 (3.46GHz, 130W), X5667 (3.06GHz, 95W), E5640 (2.66GHz, 80W), E5630 (2.53Hz, 80W), E5620 (2.40GHz, 80W), L5630 (2.13GHz, 40W), L5609 (1.86GHz, 40W). All of those chips, with the exception of the Xeon L5609, have eight threads, Hyperthreading and Turbo Boost.

 

[cbn]

Cache helps a TON for overclocking, every architecture has a point where increasing MHz won't do anything because its memory system cannot keep up with the increased power from the cores. While that specific threshold is way out of reach of even the most extreme overclocker (lets say 10-15ghz for Nehalem) the closer you get to that point the less linear of a performance gain you get. Adding more cache pushes that boundary up higher. Most of my graphs are complete bs, but this one is actually meant to be helpful.

Below we have the scaling on two processors, they are completely identical on identical systems. The only different between the two is the red line has 1MB L2$, while the green line has 3MB L2$. Notice at stock speeds, 2.5ghz, the difference isn't really that much, but as speeds are increased the 1MB L2$ starts lagging behind while the 3MB scales a lot better

Yep, these 32nm Clarkdales no doubt use less energy/make less noise at the same clockspeed compared to two cores from a Core i5 750. Trouble is it seems the 32nm cores need more clock speed than the 45nm to do the same work per unit time.

At least we found out hyperthreading works well for games under Windows 7. Before Clarkdale there was no way to test the effectiveness of this AFAIK.

 

[Voo]

Why does ATI/Nividia update drivers and whats the benefit of those upgrades, Does i5 have have a IGP ondie or not ? As for the FTC I think it best for self I stay ontopic here .

Ah you found a new back door.. congratulations! Oh so the whole talk about the performance differences between i5-750 and 661 it's all about the IGP, oh how could I oversee that? In this case you are completly right, the 661 is WAY better than the 750.

Sigh, doesn't it get boring after some time, to claim something, get called on it and just talk about something completly different so you don't have to admit that you have no clue?

And if your claims are true the whole FTC stuff is fully on topic.. if not you just invented something again - don't worry the mods are used to it.


@ExarKun333: So far I haven't heard anything about 32nm quadcores but the Xeon ones (that's nothing new), so he surely meant consumer grade 32nm quadcores where it looks rather dark. All of the listed processors have the Xeon nomenclature. Though you could argue if the only difference is in the name (how expansive will the Xeons be?)

 

[edplayer]

Could you get it through your head i don't care what you think? It's my money and i'll do whatever i want with it. The chip is an absolute ripoff yet i don't care.

you are kind of new here so you may not be familiar with all the nuts we have.


Make a new thread when you get your 670. Overclocking, benchmarks, photos everything...

It will be a lot more entertaining than about half the posts here.

 

[toyota]

you are kind of new here so you may not be familiar with all the nuts we have.


Make a new thread when you get your 670. Overclocking, benchmarks, photos everything...

It will be a lot more entertaining than about half the posts here.

because buying a chip that cost 100 bucks more and does less is not being a nut? when every review site has made it clear that for enthusiasts/gamers even the 660 is a stupid choice compared to the 750 then who in their right mind would spend even more for the 670. he did start a thread and it looks like he didnt get a 670 anyway so kudos for him for not being that dumb.