i7 5820k vs 4790k

Toggle sidebar Toggle sidebar
Status
Not open for further replies.
F

fnaan

Distinguished
Jul 4, 2014
8
0
18,510
Hi

I'm planning to buy the new Hardware computer ( x99 , ddr4 , GTX 8xx/9xx )

Anyway , is there big difference between 4790k and 5820k and dose it worth the money ?

Actually I'm disappointment for the 3.3 GHZ to i7 5820k because the i7 4790k has 4.4 GHZ 🙁

So dose it worth the money ? Or I just go back with 4790k and the old hardware ( z97 , ddr3 ) ??
 
Solution
JohnnyLucky
Just in case you did not know it, you would have to buy the new CPU, a brand new X99 motherboard, and brand new DDR4 memory. You cannot use the new cpu with older boards and older DDR3 memory. Right now you can expect to pay $1,000.00 or more for those three components.

I read numerous technical reviews comparing the Intel Core i7 4790K with the new cpus. The 4790K is a definitely better value and it costs a lot less.
Sort by date Sort by votes
Just in case you did not know it, you would have to buy the new CPU, a brand new X99 motherboard, and brand new DDR4 memory. You cannot use the new cpu with older boards and older DDR3 memory. Right now you can expect to pay $1,000.00 or more for those three components.

I read numerous technical reviews comparing the Intel Core i7 4790K with the new cpus. The 4790K is a definitely better value and it costs a lot less.
 
Upvote 0 Downvote
Solution
When looking at prices the i7 5820K looks very appealing because it is only $50 more than the i7 4790K and you get double the PCIE lanes, and six cores for just an extra $50.

However, what you might not realize is that your extra $50 on that processor will actually become more than $300 extra on the entire system. Because you need a X99 motherboard which starts at $250, and you will need DDR4 RAM which starts with quad channel kits of around $180-200.

x99 motherboard, plus ddr4 ram will cost around $450-$500 alone. Where as a Z97 motherboard and DDr3 Dual channel kit costs around $250-$300.

So it depends on what you do, if you have multiple graphics cards or heavy CPU programs then a X99 system will easily be worth the money. But if not then a mainstream Z97 system will be a lot better price to performance ratio.
 
Upvote 3 Downvote
If you're planning on overclocking at all the 5820k is worth it.

the 4790k is pretty nice, but even with the best cooling, if you push the chip, it will downthrottle to 3.7GHz because of the poor thermal interface between the silicon and the heat spreader. The heat gets trapped between the chip and the heatspreader, causing thermal throttling. Oh, it will show up as going 4.7GHz pretty easily. It will even boot up at that speed. It just won't run that fast for more than 2 seconds. It gets up past 90C for a moment, then throttles down to 3.7GHz. I know this in particular because I have one. It's now my secondary because I went back to my 3930k, which runs at 4.5GHz without throttling, for my main. The 4790k was extremely disappointing.

The 5820k is better mainly because of the interface between the silicon and the heatspreader. It's not the full solder of the SB-E or the IB-E, but it is a very good epoxy. From what I've read, it is as good as the solder interface of the old chips.

Here's a kick in the teeth, too. The X99 would work with the SB-E and IB-E chips, if Intel had allowed motherboard manufacturers to use it. The Haswell-E uses the exact same interface to the chipset as the SB-E and IB-E. Also. the Haswell uses the same interface as the SB and IB. So, the older chips could work with the newer chipsets, but Intel artificially limits the interface to prevent this through bios and firmware.

For those pushing overclocking the 4790k, run CPU-Z, then run a burn-in test, and tell us what you see. Does it really stay at your claimed overclock? (Well, those who de-lid the chips can get theirs to do so.)
 
Upvote 0 Downvote
@dgingeri

You just ***** up. Cooling on 4790k is rather easy, as even my $40 scythe mugen 4 can cool 4.7ghz under 75c. Throttling will not occur before 102c. Chances of the 4790k going to 4.6-4.8 are very, very high. If you want to go higher than that, you just need very good cooling. But you already need exceptional cooling to push the 5960x to 4.4-4.5ghz.
 
Upvote 0 Downvote


Your post is my exact same situation now! I'm leaning strongly towards the 5820k setup. Since I'm not a heavy gamer and will only have 1 GPU I plan to take advantage of the extra 2 cores setting up VMs for development, server and networking labs. I have priced both setups with a single GPU and the X99 system is approx $300 more with the higher priced MB and DDR4. This assumes I pickup a X99 board for approx $250. The CPUs are about $50 diff in price, which seems amazing for a 6 core. $300 more doesn't seem that much for a system that will easily last 4-5+ years. Plus over the years the X99 will allow for more upgrades, including possibly a CPU refresh and large amounts of RAM. Even with all that, I have never had a cpu at home where task manager shows 12 cpus:)
 
Upvote 0 Downvote


That's what you think. Due to the poor thermal interface between the silicon and the IHS, heat gets trapped in the chip, no matter how good your cooling system is. Why do you think Devil's Canyon used a modified thermal compound? It still isn't good enough, though.

If you have a Haswell, try it out. Run something to monitor the actual clock rate, like CPU-z. Run something that shows the temperature and history, if you can. Asus's AI program does pretty well with that. Finally, run something that caps out the processor at 100% and still allows your other programs to show on screen. Watch the clock rate and temps carefully.
 
Upvote 0 Downvote
Do you want me to upload a 12 hour video of prime95 with my cpu on 4.7ghz with hwinfo64, coretemp and cpu-z open? Even before remounting my cooler and dropping almost 10c in temperatures a 10 hour prime test hwinfo logfile didn't show a single drop in frequency nor any occuring throttling.

Honestly, you're the only person to ever claim that you can't overclock haswells or especially the 4790k. There are countless tests and reviews out, with cheap or expensive cooling solutions and not a single one says 'we were unable to overclock'.
 
Upvote 1 Downvote


You're correct that its thermal performance is still mediocre, but based on the body of reviews and benchmarks on the topic I think you're overstating things a bit. I suspect your attempts at cooling are not as good as you think they are. Besides, a single sample of a chip doesn't mean a whole lot, especially when Haswell is so variable in its potential.
 
Upvote 0 Downvote


This is actually a really old issue:

http://www.extremetech.com/computing/157337-the-haswell-paradox-the-best-cpu-in-the-world-unless-youre-a-pc-enthusiast

"Ivy Bridge and Haswell are both based on Intel’s 22nm FinFET process, but the integrated voltage regulator exacerbates the issue. It also seems that Haswell’s built-in thermal throttling is much more aggressive than Ivy Bridge: Where the Core i7-3770K is happy to sit at 3.7GHz under full load at 90C, the Core i7-4770K throttles back to 3.5GHz within moments of starting Prime95."

Plus people have seen different behavior from the thermally throttling:

http://www.overclock.net/t/1398276/haswell-ocing-max-temp-limits

6th post: "Mine goes up to 95c and never throttles. This is during prime95. Using HWinfo64 and never seen the minclock change. The chip likes 95c no matter what air cooler I used."

In my case, I had both of my Haswells throttle at 90C, and hung right around 80-85C while throttled down to 3.7GHz (4790k) or 3.4GHz (4770k), both being below stock clocks. Even completely not overclocking them, I would get thermally throttling down to these levels. The 4790k is supposed to run 4-4.4GHz, yet mine throttles back to 3.7.

Don't think it's the cooler or a bad install of the cooler, either. It's really difficult to mess up the install of a Corsair H100i. When taking it off, the TIM, Arctic Cooling's MX-4, was evenly spread and looked like the cooler was evenly and properly installed. I'd take pictures, but I've long since given up on that platform.

The 4770k broke when I tried to de-lid the chip to improve the thermals. The 4790k is now installed in a VM host, on an H97 mATX board and using the retail cooler, I'm using for working on my VMWare certs and learning how to use many aspects of Linux.

Maybe I just got two straight poorly built chips. Maybe the adhesive for the IHS was not properly pressed down, leaving too much of a gap between the chip and the IHS on both chips. That would be a coincidence, and I don't believe in coincidences.

Maybe I am just still upset that along with these chip problems I also had the same issue with three Asus Z97-WS motherboards, and after the third they refused to grant me another RMA, even though I got the same problem of the second memory channel not working with 3 CPUs, 4 memory kits, 2 power supplies, and 2 cases. Maybe I'm just upset that I spent over thirteen hundred freaking dollars on that project and ended up with nothing working in the process. Yes, maybe I hate the whole line just because of this project that started in May. Now I have a Core i7 4790k, an H97 mATX board, a broken 4770k, a Pentium G3220, a 16GB DDR3-2400 memory kit, and a non-working Z97-WS to show for it, and most of them aren't doing anything. After all that, I'm sitting back on my Core i7-3930k on a new Asus x79 Sabretooth with a Gigabyte X79-UP4 that doesn't supply power to any of the USB ports or PCIe slots, the reason I started the whole project in the first place. Maybe this SUMMER OF HELL, where nothing I tried between May and July worked, has made me a little more bitter than I realized.
 
Upvote 0 Downvote
Thats a lot of issues you have right there, but they absolutely irrelevant to the 4790k. If thermal throttling occurs at 90c (besides you even reaching 90c with a h100i means either the h100i doesn't work or wasnt mounted correctly) that's either a wrong temperature reading or your chip is DoA, the motherboard fucks up your psu can't handle loads. Along these issues, the 5820k would suffer the exact same.

However, sorry at your losses, thats more trouble with tech in one summer than I had in my whole life.
 
Upvote 1 Downvote


I've been building my own systems since 1992. I've been overclocking since the days of jumpers, pumping my 486SX from 25MHz up to 33MHz. I think I know a bit about how these things work. I now work as a systems admin in a software test lab for server software.

The H100i is working perfectly right now on my 3930k, keeping it at a nice, calm 42C max while running 4.5GHz while playing Star Trek Online, 25C at idle. I've actually had this H100i since shortly after it came out. The MX-4 I use was spread out evenly upon removal of the H100i from the Haswell chips, looking much like the pattern I get with the 3930k. It was mounted and working perfectly. It wasn't the cooler or my ability to mount it. If you know anything about how an H100i mounts, it is really, really difficult to screw it up, unless you manage to break the motherboard. The problem is definitely in the chip packaging.

Granted that the Z97-WS board may have read the temps wrong. I keep getting boards with that second memory channel not working, so who knows what else might be wrong with it. However, my experiences with other Asus products has been stellar. Also, the throttling takes place within the chip, set by Intel in the factory.

It is perfectly reasonable to think that Intel knew they made an inferior chip and programmed the firmware on those chips to throttle at a lower point because they knew the thermal interface of those chips was not as good as others, yet still within their specs. How many manufacturing managers, trying to keep costs down and profits up, came back with a "the stupid customer will never know" when it came to a certain level of defect.

I knew it right away when I overclocked my 4770k to 4.7GHz and got benchmark reports that didn't even keep up with a stock 4770k. That's when I brought up Asus AI 3 and watched the clock rates and temps of each core on a secondary screen while running a burn-in test on the main screen. They probably didn't expect users to actually test their stuff that thoroughly. Two in a row, one 4770k and one 4790k shows me that they don't have their manufacturing tolerances quite as refined as they like us to believe, and shove parts out the door that don't quite make advertised specs.

Considering the nature of the thermal interface between the 5820k and its IHS, it won't have the same problems. It basically can't, or the entire chip won't function.
 
Upvote 0 Downvote
If the chip or motherboard are defect, the difference between TIM under or a soldered iHS arr unimportant. Your motherboards or cpu's were defective, or something else was wrong. If the same thing happens with a 5820k/x99 it won't magically work. Chances that something is messed are equally small. Out of millions of haswells having two not working is not a reason to not recommend them. Same would go for Haswell-E or other architectures.
 
Upvote 0 Downvote
That's the trick. Me getting two "not quite defective" CPUs shows me that this is very likely far more widespread than most people realize. I would be willing to bet good money that a great deal of people are having the same problem I had, but are blissfully unaware they've been cheated because they don't look closely enough at how their systems perform. Many are probably too proud to admit it even if they did find this. Out of millions, the chances of getting two in a row, of two different lines, with the same near defect, (note, they still run, they just run slowly, so they aren't exactly defective) means that the numbers of this same near defect are quite likely anywhere from 20% to 60% of the supply, depending on how much of it Intel's management decides to tolerate.

My motherboard defect was the memory channel, not the throttling. That would be impossible since Intel put the throttling circuitry into the CPU itself in Haswell. So, that's out of the argument.

The difference in the way the 5820k is manufactured makes it impossible for the 5820k to even have a defect like this. If it did happen, it would leave the silicon completely out of contact with the IHS, and the chip would cease to operate entirely. If there were a larger gap, more solder/epoxy would fill in between the two and, since solder/epoxy is MUCH more efficient in transferring heat on its own compared to the Haswell thermal goop, it wouldn't create the same type of problem.

So, yes, a 5820k would have a very significant decrease in the chances of getting a chip that could not be properly cooled due to an internal Intel near defect like this. It makes the 5820k a much better chip.
 
Upvote 0 Downvote
That's absolute bullshit, honestly. 20-60%? You are not being serious, are you? It's most likely below 1% and the absolute same can happen with solder instead of TIM. And the chances of your actual cpu being defective are still small as well, chances your motherboard was defect or badly calibrated the cpu voltage/fan speed are much higher. You didn't even take the chip into the second system to test it.
Seriously, blaming an entire cpu architecture on bad, bad luck or own/other failure is just... retarded.
 
Upvote 0 Downvote


I disagree with you, my results were completely different when doing similar testing...
You're not an intel engineer and I suggest you don't start making conclusions based on a couple samples.
"I suggest you read the specifications from intel"

BTW I suggest you compare the you read about t case temperature and tj max...


Tcase Max is the maximum temperature that the Tcase sensor should reach. Both Tcase and the thermal specification information can be found on the Intel web site.

Did you know the t case max is higher on has well?

Tjunction Max is the maximum temperature the cores can reach before thermal throttling is activated. Thermal throttling happens when the processor exceeds the maximum temperature. The processor shuts itself off in order to prevent permanent damage.
 
Upvote 1 Downvote
Wrong. The Haswell-E processors overclock MUCH easier than any other haswell processor. The reason is that they use solder between the chip and the heatspreader instead of paste. Most haswell-E processors can use a board's auto overclocking feature and get to 4.5+ easy.

If you're more productivity based, get the 5820k and overclock it. You won't be dissappointed.
For gaming, I recommend AMD. I simply can't justify the price for a couple extra fps. Just my opinion, not trying to start a flame war.

 
Upvote 1 Downvote
@dgingeri - You really need to remember that not every chip is the exact same. So where you are getting your info is more than likely mislead or your chip is screwed up. I have a 4790K OC'd to 4.7Ghz and there is no throttling. My temps max out at about 81'C and its 100% stable. If you have more questions about my setup then look at my profile. More than likely you have a crappy chip and you did overstate a lot about this chip.
 
Upvote 0 Downvote


As I'm sure you've been told in other replies it depends what your using it for. If for compute tasks go with Haswell-e, If for gaming the next thing to take into account is whether your going to want to use multiple graphics cards with it. If your not going to use more than 2 graphics cards then the 4790k is the best option. As well as being a cheaper build overall it will outperform Haswell-e a lot of the time, However if your considering 3 or 4 graphics cards then you need to go with Haswell-e but take note of the fact that the 5820k has not got a full pcie lane count like the other two models. You'll need to consider things like how Nvidia sli requires x8 (pcie lanes) per card. I think AMD requires x4 but don't quote me on that, my memory may not be exactly right, Research it yourself to make 100% sure. Remember things like how it's no good getting a 5820k if you want to go with a 4-way Nvidia set up, To do that you'll need one of the other two Haswell-e chips.
 
Upvote 0 Downvote
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom