Intel "Devils Canyon" Launch - Reviews, OC Info, Availability

[darb]

According to ARK, VT-D and TSX are present on the 4790K, so all you lose is TXT and vPro.

I thought VT-D and overclocking were limited to socket 2011. Thanks for the info! Now if only the 4790K supported ECC memory, I could build my ideal VM Hypervisor / FreeNAS / gaming rig.

 

[tidy]

.

 

[rtsurfer]

And as I said in the other thread, in reality, it's probably not even going to run at 4.4 with 'one core loaded'. In my experience, turbo is unpredictable and next to useless. On paper, the turbo steps look nice and clean and logical, but it doesn't operate that way in practice.

Yup.
I just felt the need to point it out since all the claims of "Intel Validated 4.0/4.4" around here.

Shouldn't the 4.4 on single core hold as long as you are running a decent cooler? That is, not the stock one.

 

[Galatian]

Yup.
I just felt the need to point it out since all the claims of "Intel Validated 4.0/4.4" around here.

Shouldn't the 4.4 on single core hold as long as you are running a decent cooler? That is, not the stock one.

It usually does. I mean I'm only speaking out of personal experience but when I have a heavy task my Intel CPU does go to the set turbo clocks. Not sure what he meant with wild fluctuations?!?

 

[tidy]

.

 

[Enigmoid]

It's been the same on my 750, 2600k, 3770k, and 4771. I don't have access to a turbo CPU right now (sold for upgrade), but I found an old video I took of UT2004 running on the 2600k that demonstrates classic turbo behaviour (please excuse the quality):

http://youtu.be/KLebLUsLVG0

The 2600k's turbo steps are 3.8GHz for one core, 3.7 for two, 3.6 for three and 3.5 for all four. UT2004 is a single threaded program, and there are no other programs running. You don't even see 3.8 GHz at all. Well maybe I can give 'turbo' the benefit of the doubt and say that driver overhead is putting a (tiny) load on another core. It should then at least be running at a stable 3.7GHz. But no, it spends half it's time at 3.6GHz, constantly changing, and even dropping to the lowest 3.5GHz turbo state. What good is that? The game is single threaded. The temps and TDP are nowhere near max. Maybe if you set up a completely synthetic test like Prime95 on one thread turbo might 'work', but that's no good to anyone.

The concept makes sense, and if it worked properly in real-usage situations it would be great (especially for 6-8 core processors), but my experience with it makes me think 'turbo' is barely any less of a stupid gimmick than 'GPU boost' (a truly pointless feature if ever I saw one) .

Very rarely are 1 core turbos applied. 2 cores should be seen. Might be a motherboard/bios thing.

 

[BFG10K]

http://www.legitreviews.com/intel-core-i7-4790k-devils-canyon-processor-review_143880

http://semiaccurate.com/2014/06/10/intels-core-i7-4790k-devils-canyon/

Given that the 4790K is 4.0/4.4 boost on stock settings, would that bring about silent data corruption? Or is there no guarantee of that?

No, why would it? 4.0/4.4 are official speeds from Intel - they're stock clocks, guaranteed from the factory.

What we're talking about is some internet guy going into the BIOS and raising the clock and/or voltage outside of default settings.

Also silent data corruption is just one possible problem with overclocking. I specifically highlighted it as an example because Prime/Linpak (et al) can't test for it.

I suppose the safest option for gamers who also store important files on their machines would be a non K SKU, or a Xeon like the 1231 V3.

Huh?

 

[moonbogg]

Chip is good, but not overly exciting. When overclocked, its faster than a stock 3960x. That's cool.

 

[Techhog]

Given that the 4790K is 4.0/4.4 boost on stock settings, would that bring about silent data corruption? Or is there no guarantee of that?

I suppose the safest option for gamers who also store important files on their machines would be a non K SKU, or a Xeon like the 1231 V3.

So, you think Intel is so stupid that they just pick random settings for their CPUs without any testing for stuff like this? Because if that's the case, then all of their CPUs are at risk, K or not.

 

[Techhog]

There is is again!

Absolution does not use 6 cores. The difference between 3 and 4 cores was negligible - I tested it myself. So what the hell is going on here?

I'd all but decided on DC, but every time I see something like this it steers me back towards Haswell-E. There is clearly some potential gaming advantage to HEDT that has nothing to do with the cores. I'd really like to see a comparison with downclocked memory, or run it in dual/single channel to see if it has anything to do with that.

It's probably just that hyperthreading hurts performance.

 

[mistersprinkles]

One opened near me recently. If only I had money...

I doubt that my board has the power delivery for it anyway. :/

Your board doesn't have the chipset for it.

 

[Techhog]

Your board doesn't have the chipset for it.

Z87 boards can support it. It's a matter of getting BIOS updates and possibly power delivery. Several 8-series boards have been updated for it.

 

[witeken]

FPiednoel is giving a helpful OC tutorial of DC on Twitter: https://twitter.com/FPiednoel .

 

[StinkyPinky]

FPiednoel is giving a helpful OC tutorial of DC on Twitter: https://twitter.com/FPiednoel .

If only we could actually buy the CPU's.....

 

[RaistlinZ]

It seems like 4.7Ghz is the average most are getting from the review sites. Pretty nice, but nothing earth shattering.

Then again, it is the same silicon as the 4770k so I'm not sure why people had such wild expectations to begin with.

 

[notty22]

Anyone getting a Anniversary Pentium ? Does anyone know whether this cpu is getting these fetures?

I ordered one. Had the original Pentium also



edit: Here is a quick o/c review.

http://www.digitalstormonline.com/u...-review-and-overclocking-benchmarks-idnum298/

For our tests, we used the ASUS Z97 Deluxe motherboard coupled with 16 GB of Corsairs Dominator Platinum DIMMs at 1877 Mhz running in Dual-Channel. All tests were done with the stock Intel heat sink, even the overclock benchmarks. When we put the Pentium chip under water and tried to push it further, it would BSOD regardless of the voltage we threw at it so we felt just including the stock heatsink numbers would be enough. After all, this is a budget chip so the numbers we got will likely be achievable by most users.
- See more at: http://www.digitalstormonline.com/u...ing-benchmarks-idnum298/#sthash.9s6BxwnF.dpuf

​

 

[toyota]

There is is again!

Absolution does not use 6 cores. The difference between 3 and 4 cores was negligible - I tested it myself. So what the hell is going on here?

I'd all but decided on DC, but every time I see something like this it steers me back towards Haswell-E. There is clearly some potential gaming advantage to HEDT that has nothing to do with the cores. I'd really like to see a comparison with downclocked memory, or run it in dual/single channel to see if it has anything to do with that.

yes those numbers are BS. you can look at any other cpu review for that game and see that 4770k does way better than that. even my minimum framerate is over 70 fps in the bench.

see here the 4770k and even 4960x are dead even. http://www.xbitlabs.com/articles/cpu/display/core-i7-4960x-4930k-4820k_6.html#sect0

see here the 3770k and 3960x are dead even. http://www.techspot.com/review/608-hitman-absolution-performance-benchmarks/page6.html

 

[toyota]

It's probably just that hyperthreading hurts performance.

thats not whats going on there. the game does not make any real use beyond 4 threads or 4 cores but HT on 4 cores would not slow it down like that. that bench is bs or they have some driver issues.

 

[Pheesh]

jesus christ twitter is an awful medium for things like this....reading explanations 140 characters at a time

 

[tidy]

.

 

[toyota]

False, this is not the first time this effect has shown up in this same game (not to mention some other games), as I've posted before:

http://www.maximumpc.com/ivy_bridge_e_review_2013?page=0,3

that is NOTHING like that earlier bench showing a nearly 50% advantage for the 6 core cpus.

 

[tidy]

.

 

[toyota]

You're pulling 'nearly 50%' out of your arse, but sure, the effect is not as great. However, it's still significant, and now two separate reviews of the same game have shown it to some degree. I'm not inclined to ignore that completely.

What's to say that certain parts of the game aren't more cache (for example) dependant than others? The reviewers aren't all following the exact same benchmark procedures.

Also:



What bench? Depending on the area, Absolution is ridiculously CPU bound. This is with a 4771 at 3.5GHz:

]http://i.imgur.com/St74r8S.png

I doubt you're playing the entire game without frequently seeing sub70 frame rates.

perhaps you should pay more attention before telling me I am pulling stuff out my arse. 46.1 to 66.9 is 45% so yeah "nearly" 50% just like I said. did I say the whole game was over 70? NO, I said the benchmark. and what benchmark do you think I am talking about? the built in Hitman Benchmark of course. and you must be on old drivers because the last 2 sets of driver made a massive cpu bound difference and I am over 60 fps in the same spot you are showing. I know all about this game as I complained before about the cpu limited parts but again they have been addressed on the last 2 drivers.

 

[StinkyPinky]

False, this is not the first time this effect has shown up in this same game (not to mention some other games), as I've posted before:

http://www.maximumpc.com/ivy_bridge_e_review_2013?page=0,3

Sp 5 fps difference instead of 21 fps difference. That's not even in the same ballpark.

 

[tidy]

.