Intel Core i7-3820 Reviews

[onethreehill]

First Review?
@Tomshardware
http://www.tomshardware.com/reviews/...mark,3090.html

[Spectrobozo]

interesting test, the 3820 looked better than what I expected (in terms of cost and performance),
I think it may be a better choice than the 2600k, if the price is correct,
the platform cost is not to high, there are 220 usd x79 MBs on newegg, while there are many z68 at around 200 to or more (I know I'm probably comparing a more robust z68 against a not so much x79, but I think what you gain in terms of performance and upgrade possibilities worth it)...
the bclk OC stability issues are worrying, but I suppose they can improve on that in future bios? (since games showed more problems, maybe there is something wrong with the pcie clock?), quad channel costs more, but if you don't want it you still can use dual I suppose, and memory is cheap anyway...
so probably the biggest con is power consumption, but I don't think is to much of a problem for someone buying and overclocking the i7,
also one very positive thing, you can upgrade later to a 6-8 core CPU, while you are stuck to 4 with the lga 1155.

[El Mano]

The review is absolutely retarded. How can they test a CPU using a 1920x1080 res with one GPU?
It's pointless.
Besides they used games that demand much more GPU power than CPU.

[PatRaceTin]

most interesting socket 2011 cpu

[3NZ0]

1 gpu?

1 gpu?

Was that review conducted with only 1 gpu? Goodness me some people....... I'd expect a minimum of 2 580/6970's, possibly even 3 or a pair of 590/6990's thrown in as well.

Outside of productivity that review is so badly gpu limited it hurts to read it.

[kl0012]

http://www.anandtech.com/show/5276/i...sandy-bridge-e

[Lanek]

The test of Toms hardware is really unreadable .... im sorry, but they put in red the OC version... i like they do a test of OC vs OC, but you dont review a processors with a 1Ghz OC and speak only about that: If it was against a 2600-2700K OC, peoples will have a better idea instead of place it against the 3960K... Then you have nearly no one words about the stock 3820. They really want to sell you this CPU lol.

Anandtech have do a better test... for sure.

[iboomalot]

should drop the rez to 800x600 to take the GPU out

[bamtan2]

http://www.anandtech.com/show/5276/i...sandy-bridge-e

the is the real news, not the toms one.

the commenters and reviewers dont understand these chips because they aren't workstation or server builders. even anand doesnt get it. his conclusion is that this is good if you need more than 32gb of memory, but if you try to put just 32gb of memory in a 2600k system, it will cost more than the 3820 system. only the 3820 can use 2x more cheap 4gb dimms. not to mention unbuffered ECC dimms, which should have been mandatory years ago...

[Lanek]

completely agree with you... Well at least for me there's another reason for buy it: - can build a good base SB-E ( good board and ram ), with a cheap processor, before sell it for a 3930-3960.

[LightSpeed]

Nice CPU, only draws slightly more power than the 2600K (thanks to a non-harvested die), slightly faster and slightly cheaper as well so im wondering if I should sell the current 1155 for a 2011 but with IVB coming out for 1155 first, its a tough call.

[paulbagz]

People are raging around the review but keep overlooking the fact that its Toms Hardware lol.

There's a reason why Toms reviews are .

-PB

[R101]

People are raging around the review but keep overlooking the fact that its Toms Hardware lol.

There's a reason why Toms reviews are .

They are a hit and miss. Mostly miss.

[BababooeyHTJ]

The Anadtech review is up.

I have to admit that I was disappointed not to see Fallout 3 or Skyrim benched.

[K404]

If all the benches are GPU-bound, does that not suggest that this chip is a good purchase versus, say, the 3960X and a user can save their money for no gaming compromise?




Test it that way, call it excellent value for money, everyone wins, no lie is told.


P.S... not defending any review site.

[bamtan2]

If all the benches are GPU-bound, does that not suggest that this chip is a good purchase versus, say, the 3960X

if there is one guy this confused, there are others. PROCESS THIS: sandy bridge-e for x79 is just sandy bridge with more professional features that have nothing to do with gaming. the benchmarks are not ambiguous. in games these chips perform EXACTLY like last year's sandy bridge.

[BababooeyHTJ]

if there is one guy this confused, there are others. PROCESS THIS: sandy bridge-e for x79 is just sandy bridge with more professional features that have nothing to do with gaming. the benchmarks are not ambiguous. in games these chips perform EXACTLY like last year's sandy bridge.

Except in games that seem to scale from the extra cache like Fallout 3, Skyrim, and Wow. Look at the tom's hardware Sandybridge review, the 2600k performs much better than the 2500k than the mild clock speed bump could explain and lord knows that it wasn't the ht. The anand wow bench in the sandy review was similar. Same thing with the Skyim benchmarks that I've seen with Sandybridge-e, I think hardware heaven did one. There are more out there if you want to look. Just because something like Metro 2033 doesn't perform and differently doesn't mean that you can make any sweeping generalizations. Not to mention that you can bump up the base clock with sandy-e, that could offer more performance as well. I haven't seen any solid benchmarks on that. I will say this. The games that I have tested seemed to scale better with overclocking on lynnfield that I saw with sandybridge. Could base clock overclocking have anything to do with that? Its hard to say for sure without some solid benchmarking practices in games that scale with cpu speed like Skyrim.

[K404]

if there is one guy this confused, there are others. PROCESS THIS: sandy bridge-e for x79 is just sandy bridge with more professional features that have nothing to do with gaming. the benchmarks are not ambiguous. in games these chips perform EXACTLY like last year's sandy bridge.

Sure, the info could be interpreted the other way, but what I said is still accurate, call it "reviewer spin"

[bamtan2]

Just because something like Metro 2033 doesn't perform and differently doesn't mean that you can make any sweeping generalizations.

the performance advantage over 2600k is zero to 5%, and the 3820 is clocked faster. so the architecture does next to nothing. numbers: http://www.anandtech.com/show/5276/i...ndy-bridge-e/3

anand concludes the same thing.

[BababooeyHTJ]

the performance advantage over 2600k is zero to 5%, and the 3820 is clocked faster. so the architecture does next to nothing. numbers: http://www.anandtech.com/show/5276/i...ndy-bridge-e/3

anand concludes the same thing.

Did you read the rest of my post? You can't make sweeping generalizations from a couple of timedemos with a single 5870.

Here is Skyrim in the hardware heaven review.

Check out shogun 2 and arma 2 in the bit-tech sandy-e review.

Here is the anandtech 2600k review with a fallout 3 benchmark which combined with benchmarks like that hardware heaven skyrim benchmark leads me to believe that Gamebyro likes the cache.

If you dig around you'll find some other surprising results.

[sin0822]

The review is absolutely retarded. How can they test a CPU using a 1920x1080 res with one GPU?
It's pointless.
Besides they used games that demand much more GPU power than CPU.

1 gpu?

1 gpu?

Was that review conducted with only 1 gpu? Goodness me some people....... I'd expect a minimum of 2 580/6970's, possibly even 3 or a pair of 590/6990's thrown in as well.

Outside of productivity that review is so badly gpu limited it hurts to read it.

You guys understand this isn't a gaming review, its a CPU review, right? Sure the CPU might be bottle-necked, but if they were to change their setup for this platform, then they would have to go one every other previous platform and change the GPU.

Did hardwareheaven finally give out an award? lol

Different games take advantage of different things, just because a CPU does well in some games, doesn't mean it is a gaming CPU. It all depends on what you do with the CPU. It is hard to make generalizations unless results are consistent across the board, or at least 90% of the time it wins by some margin already calculated to be sufficient. it is like the 7970.

[bamtan2]

Did you read the rest of my post? You can't make sweeping generalizations from a couple of timedemos with a single 5870.

Here is Skyrim in the hardware heaven review.

Check out shogun 2 and arma 2 in the bit-tech sandy-e review.

man you're in the wrong thread. 3960x has 6 cores and is a different design. we are in a 3820 thread. 4 cores.

3820 performs just like 2600k/2700k in games, just like the benchmarks show.

[herderien]

it's the same architecture, just less cache and cores, or am i mistaken ?