Intel Core i7 4820K and 4960X Ivy Bridge-E processor review

It's a bit disappointing but not unexpected, it was the same thing with 2600k - 3770k. I'm really looking forward towards haswell-e and beyond, it's about time that the industry pushes for raw performance again,

 

I'm kind of suspicious of the RAM throughput readings in AIDA.

I will do some testing tonight with my 3820. It really looks to me like the 3820 on that chart was running in dual channel mode and not quad channel mode.

If the RAM controller is really that much better, then anything that uses a lot of RAM should see a huge jump in performance.

 

Yep, I am officially calling shens on the i7-3820 figures on your charts.

Here are some quick AIDA tests I did on my system just now.

CPU speed and RAM speeds the same as tested on the 4820k




Now CPU speed at my normal OC as well as different RAM speeds





EDIT:
I bet I know what is going on. After thinking about it, I remember that AIDA/Everest used to give really low RAM scores when compared to Sandra. I am guessing that the older tests including the i7-3820 were run on an older version.

If that is the case, then those older scores either need to be revisited or all together removed from the charts since they are no longer accurate.

In either case, IB-E is not really faster in RAM throughput than SB-E.

 

No but it's stagnating to the point that by the time this 2600K breaks we still haven't reached a decent increment.

When I upgrade I like it to be worthwhile.

 

This man has a good point

Though coming from Piledriver and Deneb it will be worthwhile which ever rout i take.

 

First off; isn't it strange how we're in a thread discussing Hilbert's review on Guru3D yet instead u post bench results from AnandTech....really funny that . But not really that funny.

Maybe Anand can shed some light on this?

Here's what he has to say regarding your carefully selected benchmark;

Tbh I'm surprised you didn't post another Skyrim result as well.


Second; the 4960K is 50% faster in Monte Carlo Sim, not 100%. That translates to 6 seconds.

Right now an 8350 on NewEgg costs ~$180. The 4960K MSRP is $990, that's a difference of $800. Which equates to 500%.....for what - 6 secs and a handful of fps?

Now I'm no mathematician but even I can see those numbers don't add up.

The amount of AMD bashing here is pathetic.

 

Yep.

 

^Wow, another person who skipped math class. :stewpid:

 

If a CPU can do twice as much for the same amount of time, its called a 100% increase.

CPU A - completes a test in 12.6 sec
CPU B - completes ( a test in 6.3 sec ) 2 tests in 12.6 = does twice as much for some amount of time = 100% increase.

Maybe that's easier for you to understand.

 

The reason I chose Anandtech's power graph was because it featured the FX-8350, Hilbert's doesn't. In any case, Hilbert's power graph is within 2W of the results Anand got.

Second, take your pick of nearly any other benchmark. The 4960x is between 30%-50% faster than AMD's top of the line chip and uses nearly the same amount of power doing so. Price is not a concern for people that buy chips like these, productivity is. The fact that you can get the same amount of work done in half the time is money to people who work in such fields and with such products. $990 is but a mere drop in the bucket for professionals, they don't even take a second look at the price. I know, I've dealt with them. They just literally max out a computer build's option and order it and use it for a year or two.

Again, you're thinking small scale as a gamer; which is why you don't understand how the market functions and how Intel is able to sell chips like these at $990.

 

Your both idiots. I'm an Engineer this is my job.

If you don't believe me look at the benchmark; Anand already commented the 4770 which completed the bench in 9 secs is 33% faster than 4960k that took 6 secs.

Another 3 secs is 12 secs THAT DOES NOT EQUAL 100%.

I understand perfectly, you don't.

8350 completed the bench in 12 secs, Intel did it in 6 secs which is HALF the time AMD took. That is 50% quicker not 100%.

100% would be 0.00 seconds. Is this starting to make sense?

Seems pointless trying to tell you guys.

 

LOL, you are actually right there. I took the numbers the wrong way (from 6 to 12 instead of 12 to 6). My math teacher would whip me right now for such a comical mistake. I stand corrected there.

The rest of the post still stands.

 

Yay finally! :banana: Well at least you can admit it.

Anyway I don't agree with the rest of your post but you're entitled to your opinion, so lets leave it there before this erupts into a flame war....

 

But you didn't. That chart is comparing time...
Finishing a job in half the time ( 50% less time ) means it can do 2x, or 100%, more for the same amount of time.

6s is 33% less time than 9s. But 9s is 50% more time than 6s. So for 9s 4960x will able to complete 1.5 test's ( 50% more ) than 4770.

@

No, half the time is 100% quicker. Completing the test in half the time, 6.3s = completing 2 tests ( 100% more ) in 12.6s.

Finishing a job for 0.00 sec, would mean It's infinitely faster than anything governed by the physical laws of this universe. Not 100% faster. Everything takes time, 100% less time ( take all the time away ) is instant, or infinitely fast. Thus, you just presented yourself the reason why your way of looking at the numbers is completely flawed.

Pill, back to elementary bro.

 

But that's work, not comparative performance which I originally said.

6.3 seconds is half the time of 12.6, which is 50% (zero is the base). But if we compare work, then I can do twice as much work 6.3x2=12.6 (6.3 is the base now).

That's where I went wrong the first time around, I went the wrong way with the numbers for performance.

 

^Sure 6.3s is half of 12.6s, which is 50% less time. And that's the time 4960x needs to complete the test. Which means, It can complete twice as many ( 100% more ) tests in 12.6s. Making it twice as, or 100% faster than 8350. ( which can only complete a single test in 12.6s ).

 

Except that it didn't take 12 seconds and if it had then the 8350 would have taken 24 which means they would both still would be doing the same amount of work.

The only way Intel could do twice the amount of work is if it took half as long again as AMD did...meaning 18secs.

 

oh my god, you're clueless. I'll quote myself :

Oh wow, your edited post is even more amazing. Don't think I've read something this stupid in years. Brings back memories of stupid kids at school.
Sort of funny and sad at the same time...

 

I'm clueless, lol that's rich. No idea what you're quoting but please don't ever become an hardware tester.

Btw, tell me - do all trolls hail from Serbia or only 99.9% of them?

I'm out of this discussion. Bye Bye.

 

Don't ever become an Engineer please. Oh wait...

Yep, even here, 99.9% are just like You.