i7 5820K @4.4 slower than i5 3570K@4.4

[nyker96]

did you say 8x16GB of DDR4? damn that's more ram than i could count. but anyhoo, you'd probably have to go EEC w/ that much RAM, I'd be worried you get a RAM error here and there. ;p but damn, that's a freakin little supercomp right there!

 

[20TenGTS]

@sjerra

Even if you leave it in high performance option, the cpu still clock down on idle. It only switch between idle clock and max clock, not inbetween like balance mode. So even watching a movie will bump cpu to max clock.


You can create shortcut to the power option with the link below.
http://www.sevenforums.com/tutorials/66925-power-options-shortcut-create.html

This way you can quickly switch to high option when you need.

 

[crashtech]

Even if you leave it in high performance option, the cpu still clock down on idle.

I've seen that not to be the case, for whatever reason, on more than one occasion, that is, with 8-series chipsets. No experience with 9-series yet.

 

[toyota]

I've seen that not to be the case, for whatever reason, on more than one occasion, that is, with 8-series chipsets. No experience with 9-series yet.

yep if I select high performance it goes to 4.4 and 1.25 volts and thats where it stays until I change the power plan back.

 

[TheRyuu]

@sjerra

Even if you leave it in high performance option, the cpu still clock down on idle. It only switch between idle clock and max clock, not inbetween like balance mode. So even watching a movie will bump cpu to max clock.

I'm not sure that's right. I believe it depends on the Windows power settings under the Processor tree. Either way that doesn't matter since you really should be using balanced. Modern processors (and modern Windows versions) are good at handling power states and you really should take advantage of them.

The same goes for SMT (Hyperthreading). Unless your workload is going to be exactly like what your benchmarked where HT enabled was slower you really should just leave it enabled and let the operating system handle how it balances the load.

 

[CHADBOGA]

I'm surprised so many people thought that the first generation of DDR4 was going to be worth much, and it might in fact be a negative for performance, over highly optimised DDR3 systems.

 

[Idontcare]

I'm surprised so many people thought that the first generation of DDR4 was going to be worth much, and it might in fact be a negative for performance, over highly optimised DDR3 systems.

Was it any different with DDR->DDR2 or DDR2->DDR3? To recollection it wasn't. Getting more channels (of any kind) was a bigger bump IIRC.

I thought the value of DDR4 was supposed to be in getting same (or slightly better) performance as DDR3 but with considerably lower power consumption because the voltage has been significantly reduced.

There once was a time when Intel was willing to leave open the opportunity for the memory makers to add value to the consumer, but they got screwed with Rambus.

So instead Intel went the opposite direction. Went with beefed up pre-caching and huge on-die cache schemes to insulate their CPUs from the vagaries of the external memory schema.

And now the memory makers are left in the unenviable position of being a necessity (like a keyboard) but not really considered a key value-add in the overall performance equation.

Sucks for them, but at the same time they are the one's who wanted a business model that gouged consumers to the tune of $1000 for 128MB of RDRAM. Way to shoot yourselves in the foot dramuri.

 

[Alpha0mega]

With my 5820k, 4.0GHz at 1.1v, using High Performance power mode, the processor multiplier goes between x12 and x40 (and voltage between 0.718 and 1.1 respectively), as required. Playing just a video, as another poster mentioned, doesn't cause it to stay at max, unless it's something demanding, like 4k.

AFAIK, the only difference between High Performance and Balanced in regards to the processor, is the "Minimum processor state" setting under "Processor power management". Unless there is something else going on under the hood, that should be it.

Interestingly, I ran into some similar problems after I built my new PC. My previous PC was i7 950 with Windows 7, and the new one is 5820k with Windows 8.1. While working on the new PC, I started noticing some odd slowdowns, even in the general OS UI. Things as simple as the window minimize/restore animation started becoming jerky. The severity of the jerkiness varied. At the low end, others I showed it to for verification couldn't see it, though I could. Other times it was enough for them to see it too. When that happened, other activities were visibly effected as well, like opening programs and files, or scrolling webpages. Spent some time messing with the video card settings and drivers. when I stumbled upon the minimum state setting. Setting that to 100% fixed it instantly.

On further experimentation, I found that no matter what setting I used, as long as it was below 100%, would cause the problem, including the random severity I mentioned. 1%, 99%, didn't matter. It also effected the games I was playing, causing visible jerkiness and slowdowns. I don't know why modern, demanding games are even letting the processor downclock during gameplay.

Since then, I have set my minimum processor state to 100%, and everything has been fine. The voltage and multiplier goes between minimum and maximum only, nothing in between, but it works well, and doesn't stay at max all the time either.

 

[tweakboy]

Put everything in stock in BIOS.. Try booting up then. If you cant try safe made... let us know, also put doiwn signature so we can help you better.

 

[CHADBOGA]

Was it any different with DDR->DDR2 or DDR2->DDR3? To recollection it wasn't. Getting more channels (of any kind) was a bigger bump IIRC.

I thought the value of DDR4 was supposed to be in getting same (or slightly better) performance as DDR3 but with considerably lower power consumption because the voltage has been significantly reduced.

I think a good rule of thumb is to avoid the new memory standard till at least the second family of processors come around.

So with Haswell-E being the first family of Intel CPU's to use DDR4, it would have put me off considering it and wouldn't buy a DDR4 system before Skylake shows up.

 

[Lepton87]

With my 5820k, 4.0GHz at 1.1v, using High Performance power mode, the processor multiplier goes between x12 and x40 (and voltage between 0.718 and 1.1 respectively), as required. Playing just a video, as another poster mentioned, doesn't cause it to stay at max, unless it's something demanding, like 4k.

AFAIK, the only difference between High Performance and Balanced in regards to the processor, is the "Minimum processor state" setting under "Processor power management". Unless there is something else going on under the hood, that should be it.

Interestingly, I ran into some similar problems after I built my new PC. My previous PC was i7 950 with Windows 7, and the new one is 5820k with Windows 8.1. While working on the new PC, I started noticing some odd slowdowns, even in the general OS UI. Things as simple as the window minimize/restore animation started becoming jerky. The severity of the jerkiness varied. At the low end, others I showed it to for verification couldn't see it, though I could. Other times it was enough for them to see it too. When that happened, other activities were visibly effected as well, like opening programs and files, or scrolling webpages. Spent some time messing with the video card settings and drivers. when I stumbled upon the minimum state setting. Setting that to 100% fixed it instantly.

On further experimentation, I found that no matter what setting I used, as long as it was below 100%, would cause the problem, including the random severity I mentioned. 1%, 99%, didn't matter. It also effected the games I was playing, causing visible jerkiness and slowdowns. I don't know why modern, demanding games are even letting the processor downclock during gameplay.

Since then, I have set my minimum processor state to 100%, and everything has been fine. The voltage and multiplier goes between minimum and maximum only, nothing in between, but it works well, and doesn't stay at max all the time either.

I experienced the exact same behaviur under Win 8.1 which was installed on 2xMX200 256GB in RAID 0. When I installed Windows 7 on just one iSandForce 2 128GB the problem completely went away. I'll try your solution under Win 8.1 but I'm not sure I want my computer to draw basically twice the power at idle just because it's running Windows 8.1. Maybe I'll sit on this W7 until W10 get released. Hopefully it's going to work flawlessly with all the power saving features enabled. At first I though it might be due to that RAID but I didn't bother to check that. Is this a known problem or a very obscure one?

 

[Alpha0mega]

I'll try your solution under Win 8.1 but I'm not sure I want my computer to draw basically twice the power at idle just because it's running Windows 8.1.

Keep in mind that the processor downclocks at idle even with the minimum state at 100%. When idle, or working on some light computation, the processor goes between the minimum of x12 multiplier and 0.718v, to the maximum (of whatever your OC is).

The minimum processor state simply sets the base for the speed the processor will run at when computation power is needed. E.g. with 100% as minimum, when opening Word it will clock to x40 (in my case), then back down once the task is complete. If it was set to 50% it would have clocked to x20-25 (not sure if 0 is base or 12 in that calculation). It will not run at max while idle.

So you can set the minimum at 100% in Windows 8.1 without worrying that it will consume max power even at idle. Though, given how close Windows 10 is rumored to be, might make sense to wait. Of course, there is no guarantee that this behavior will not be present in that too.

 

[Lepton87]

Keep in mind that the processor downclocks at idle even with the minimum state at 100%. When idle, or working on some light computation, the processor goes between the minimum of x12 multiplier and 0.718v, to the maximum (of whatever your OC is).

The minimum processor state simply sets the base for the speed the processor will run at when computation power is needed. E.g. with 100% as minimum, when opening Word it will clock to x40 (in my case), then back down once the task is complete. If it was set to 50% it would have clocked to x20-25 (not sure if 0 is base or 12 in that calculation). It will not run at max while idle.
ps. I just checked. High performance option in the power plan pegs my CPU at 4.4GHz.

So you can set the minimum at 100% in Windows 8.1 without worrying that it will consume max power even at idle. Though, given how close Windows 10 is rumored to be, might make sense to wait. Of course, there is no guarantee that this behavior will not be present in that too.

Why not disable EIST in the BIOS and C-states for always consistent maximum performance, I;m going to do that and see if it makes any difference.
UPDATE: I misunderstood you I thought you meant there can also be hitching with that plan with EIST enabled.
ps. I always prefer benchmarking without EIST and C-state they can lower the result in some cases. Min and max policy is at 100%

 

[sjerra]

Dear all,

Putting the power options in high performance (or the minimum power setting at 100%) does not allow my proc to throttle back to idle. It stays at max settings. This is in Windows 7. The same goes for disabling EIST or C-states.

With my system clocked to x44 I ran prim95 with only one worker and none of the cores ever active on the thread go beyond x34. However, when I gave prime95 a core affinity (no matter which core), the active core would go up to x44.

What I'm curious about is whether this issue is limited to myself and a few others or if all i7 5xxx procs show this behaviour.
So if there are people with similar systems who could test if this that would be fantastic.

Thanks,
sjerra

CHADBOGA,
I had a need for 64+Gb of memory. Therefor my choice was limited to 8 mem slot boards:
1. Xeon ; 2. X79 ; 3. X99
Because I pay from my own pocket, the Xeon option was out. Then I compared prices and a fully kitted X79 board would be less then 15% cheaper here in europe. 15% cheaper for a system without any upgrade path (especially memory as I don't think DDR3 will go up to 16Gb sticks for none-server applications.) made me decide to go for the X99. If I had the option I would have gone DDR3 with the X99.

 

[Jorge_Orwell]

Disable all C-states and EIST in BIOS.
Set the multiplier for all cores to be the same.
Raise your uncore frequency and voltage a bit.

 

[spree47]

i am also having the same issue as "sjerra" that my 5820k just wont jump to max clock when running single threaded workloads causeing poor results in benchmarks and work including games have stutter and i watch the frequency in core temp jumping around mainly around lower frequency,but when i start up bf3 (which is a more heavily multithreaded game) it causes the frequency to be solid at my 4ghz oc, i have a 4790k htpc and it doesnt have this problem ive commented about this elsewhere and have had no luck finding someone with a 5820k willing to test but i can 100% confirm its cpu related as i had an msi x99 sli plus and now i have an x99 deluxe and it still persists

i have run super PI and its the same story when i set windows powerplan to high performance it holds at my 4ghz oc but on balanced it just jumps around 3.4ghz ish and gives a poor result, even the windows os doesnt feel like it does on my 4790k it is slower to boot from the windows logo and feels a little bit sluggish (both have 840pro)

could someone test this with a 5820k ive been wondering if its a problem with my cpu,
super pi and core temp do the best job of showing this issue that ive found

 

[spree47]

i would also like to add that when i disable 4 of the 6 cores it fixes this problem
and ppl are saying its normal cos the workload scales differently for multiple core cpus cos its a smaller percentage of load but im really just out of ideas but if its normal i wish i went with the 4790k again :/

 

[PhIlLy ChEeSe]

Hi there,
First post in over 10 years of reading Anandtech. This is how baffled I am with what happened to me.

I have an I5 3570K in a little P8Z77i with 16Gb 1600 CL8 running at 4.6Ghz with SSD. It's been running stable ever since I bought it in 2013.
I don't game. I run 1D simulation and Solidworks & Ansys FEA (Maxwell 3D). I recently ran out of memory on a job.
So I decided to buy a new rig.
After reading a lot I settled on a X99-Deluxe with an i7-5820K and 8x16 DDR4 2400 CL16 with SSD.
After some tuning I managed to run it stable at 4.4Ghz. I could probably go a little higher but I need a reliable 24x7 rig.
Anyway, when I ran the simulation I wasn't impressed at all. Ok, I didn't have memory issues, but it didn't feel as fast as my old rig. So I did a test. Both PC's side by side, same model, press start button at the same time.......and yes, my I5-3570K is faster -though after a hour it throws a memory error-.
How much faster I can' say I didn't time it, but enough for me to feel it from the start so I estimate something like 7-10%.

That's an Ivy Bridge @ 4.6Ghz vs an Haswell-E @ 4.4Ghz.
So I tuned the Ivy bridge down to 4.4Ghz and it is still faster. Now, the margin has gone down to something I'm willing to suffer (2-3%)considering the memory availability but this is very weird. I'm not expecting miracles considering this is a tock generation, but at the same clock this rig should have beaten my previous gen i5.

All my bios settings are on auto, except the memory, that's been set to xmp (so 2400 CL16) and processor clock speed. The voltage profile is set in AIsuite 3.

Is there anybody who can confirm or deny this type of behaviour especially considering one has 1600 CL8 and the other one 'only' 2400 CL16? I know from the many discussions out there that memory speed seldom makes a measurable difference. But might this be the use case where it does?
Are there any tips on why I might not get the performance I should expect?

Thanks
sjerra

LOL!
You can't compare them based on which one boots first, if so then you know nothing at all! Now if you used the same hardware then used a stop watch, I'd at least think you have a clue. Go play with your took!!