Intel: lol

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > Intel: lol

Setset: Apr 14, 2012; Grimey Drawer

redeyes posted:

140w cpus are no joke overclocked to 4Ghz:

This is using a Coolermaster Hyper 612 which is a drat large air cooler with 6 heat pipes. Never thought I'd say this, but water cooling might have been a better idea.

I just noticed that my CPU fan only runs at about 1000 RPM until the CPU gets relatively warm. You could try setting a manual CPU fan curve, maybe the fan just isn't kicking full speed in until ~80 degrees.

# ¿ Aug 21, 2016 23:32

Adbot: ADBOT LOVES YOU

# ¿ May 13, 2024 15:42

Setset: Apr 14, 2012; Grimey Drawer

feedmegin posted:

Again, cores aren't magic. Look up Amdahl's Law. You can't 'optimize towards massive amounts of cores' if the task you are attempting to complete is fundamentally sequential.

Isn't Amdahl's Law overshadowed by Gustafson's Law? Which basically states that, given a non-fixed size of data to be processed, Amdahl's law is irrelevant?

Admittedly, I dont know enough about this (because math) other than what I've read on wikis and some light research. But it felt worth mentioning

# ¿ Aug 30, 2016 15:02

Setset: Apr 14, 2012; Grimey Drawer

NihilismNow posted:

5 year old i5 2400's can deal with this fine. If anything with the move to more VDI/SBC based platforms i see corporations taking computing resources away from end users. The worker who used to have a quad core desktop with 8GB RAM now gets a VM with 2 vCPU's and maybe 5 GB RAM. Only people who can justify it get to keep their physical machines.
Or maybe i just work for horrible corporations. That's probably it.

is there a "bring an old quadcore from home" option?

# ¿ Aug 30, 2016 22:51

Setset: Apr 14, 2012; Grimey Drawer

I'd think the i7 6700 would get you better performance stock than a ~15% faster overclocked 6600k. And you wouldn't have to mess around with overclocking/higher temperatures. If the cost is the same why not i7?

Otakufag posted:

1- For those of you who have upgraded from a 2500k to a new skylake: can you feel noticeable differences when playing recent games or doing other windows stuff?
2- Should I buy a i5 6600k or a i7 6700? Both end up costing the same as the i5 requires a more expensive mobo+cooling. This is mainly for gaming btw.
3- Maybe I should wait for something better around the corner? Is kaby lake worth waiting for or something else like zen? gently caress

# ¿ Aug 31, 2016 18:04

Setset: Apr 14, 2012; Grimey Drawer

Eletriarnation posted:

I would say that a typical overclock is more like 30% over the stock base clock than 15%, (3.5 >>> ~4.4-4.6GHz) and in the absolute best case when you can load all 8 threads HT will be a 30% or so (sanity checked myself here) benefit. When you can't load more than 4 threads it gets you nothing. I don't think the non-K 6700 is a good deal for anything ever unless it's a system that can't OC.

I was thinking that since the i7 6700 has a 4ghz boost clock, then the 6600k would only get 4.6 easily so that's 15% more clocks. The i7 seems more future proof but, people gonna do what people gonna do

# ¿ Aug 31, 2016 19:35

Setset: Apr 14, 2012; Grimey Drawer

EdEddnEddy posted:

Hearing "Auto Voltage" is giving me plenty of OC related triggering. (Auto sometimes can run it just fine at "stock voltage" levels, some boards however can spike things to 1.5v+ though, also watch out for High LoadLine Calibration settings!). I guess that is to blame for the "ease of OCing" the new K series, but it sounds like few of you have really dialed into the deep parts of the bios settings that board makers like ASUS put in to really get the chips dialed in at a higher speed.

There is a ton more to do to get a stable good OC over just setting the Multiplier up and maybe the voltage and being done with it. It requires a bit more math in the Core 2 days (OC'ing with the FSB which affected the memory and such) but it is still an art to make sure you get the OC the stable voltage it needs without spiking it with voltage needlessly causing heat and possible damage long term. Of course good cooling goes a long way (I wouldn't do 4.6+ on a SandyBridge without a really good air cooler) but getting 4.4 on any of those chips I have touched is literally OC'ers Childs Play. Also of course leaving the SpeedStepping and other power saving bits in place which saves power, heat, wear, etc on the system. (Offset voltage takes a lot of time to dial in however).

It does take patience though, thats for sure. The 2500/2600K's were easy compared to the 3930K's 6Cores of crazyness. The X79 is extremely powerful, but much more, finicky.

Yeah I have no idea where the auto voltage fad came from, but back when I was first learning about OCing it was a no-no. It's just one extra setting to change with one rule, don't go over 1.x voltage. From my experience, auto voltage sets the voltage way higher than it needs to be

# ¿ Sep 1, 2016 23:04

Setset: Apr 14, 2012; Grimey Drawer

BIG HEADLINE posted:

Keeping the CPU at max speed all of the time hastens electromigration, which is a fancy way of saying that you're making current/electrons flow too quickly through the transistors, and eventually something has to give.

I don't think there is any solid evidence to support this. As long as your CPU isn't heavily - to - the - max OC'd 24/7 then it will be fine. Yeah your power bills will be higher if it stays at 4ghz all the time but it's not going to kill the chip early.

# ¿ Oct 30, 2016 20:14

Setset: Apr 14, 2012; Grimey Drawer

BIG HEADLINE posted:

I'm not really sure what Intel's thinking here: http://wccftech.com/intel-kaby-lake-core-i3-7350k-cpu/

Yay for an overclockable i3, sure...but an unlocked dual core with Hyperthreading...when for $50 more (and perhaps $30 more in a month) you could have a 6600K with four physical cores. ~idgi~

dual cores should be able to reach higher overclocks. so if someone's goal is to just have super high single-threaded performance then it might be the way to go

# ¿ Nov 13, 2016 12:59

Setset: Apr 14, 2012; Grimey Drawer

is it safe to say that we will not see any great performance gains until the 7nm graphene chips become available?

# ¿ Nov 20, 2016 18:42

Setset: Apr 14, 2012; Grimey Drawer

Gwaihir posted:

Yea, the platform improvements in 5 years have been quite extensive for storage, ram, and general I/O.

Maybe Intel just put a halt on increased performance chips so the rest of computing could catch up. Software/multi-thread improvements, ssds, higher bandwidth ram all offer better performance gains to the 5 year old 2500k than a processor with twice the IPC.

# ¿ Dec 12, 2016 17:57

Setset: Apr 14, 2012; Grimey Drawer

Pryor on Fire posted:

It's amusing how quickly the conversation has changed in the past few months. It's been what, seven? years of a steady drumbeat of "physics is hard! we hit fundamental limits! can't do any better you're just not smart enough to understand!" while Intel is just lazily minting money. Seven years of pointing out that the "physics" argument is utter horseshit and getting dogpiled/downvoted/whatever for it, or for daring to criticize Intel. They have a million PhDs, you think you know more about chip design than them? How dare you.

For these seven years If you point out the simple truth that they just haven't had any competition and that's the only reason a 6700k is barely better than a 2500k people used to lose their loving minds. No you're just dumb and don't understand physics! Do you even know what a nanometer is? It's like really tiny bro! Moore's law is dead because physics is too hard, I read it in wired it must be true.

Now it's a complete 180, anyone talking about "physics" is getting dogpiled on, and there's this blind faith that everything is going to get way better now that AMD is finally executing again.

I'm not even bitching about SA, this place is better than most other communities for sure, seems like people are a bit more on the ball and skeptical of the horseshit wired/verge/hn barfs out. On shittier sites like reddit et al people used to get banned or have comments deleted for daring to point out that Intel is just sitting on their hands, because this revealed how ignorant of "physics" they are.

I just love how we're all at war with Eurasia again and that's how it's always been, 2016 is amazing.

maybe take a break from the interwebz for a while. i think you might be conversating with yourself

# ¿ Dec 17, 2016 17:32

Setset: Apr 14, 2012; Grimey Drawer

mayodreams posted:

Linus gave us one of the best laughs last year:

https://www.youtube.com/watch?v=gSrnXgAmK8k

They may be good at benchmarking, but JFC a RAID50 striped in Windows?

god it's like watching a reality show for techies. the suspense!

# ¿ Jan 1, 2017 17:54

Setset: Apr 14, 2012; Grimey Drawer

Dali Parton posted:

Holy lol this is really absurd.

Also since we're calling out bad Youtubers, is JayzTwoCents okay? Asking for a friend.

don't listen to these guys. linus is fine if you're just a general low-mid consumer and not some wtftechlord. jayz's benchmarks are fine i just find him more annoying than some other reviewers

# ¿ Jan 1, 2017 23:36

Setset: Apr 14, 2012; Grimey Drawer

GRINDCORE MEGGIDO posted:

This guy built a giant heat exchanger underground. That's dedication:
http://www.overclock.net/t/671177/12-feet-under-1000-square-feet-of-geothermal-pc-cooling

Probably get the same results with a $100 AIO

# ¿ Jan 8, 2017 14:24

Setset: Apr 14, 2012; Grimey Drawer

BIG HEADLINE posted:

http://www.lian-li.com/en/dt_portfolio/dk-03/

but what if there is a fire and one of your arms is already filled with your cat, how will you save the computer?

# ¿ Jan 16, 2017 13:06

Setset: Apr 14, 2012; Grimey Drawer

or you could just upload to google drive and let them do all the backups for you

# ¿ Jan 19, 2017 13:41

Setset: Apr 14, 2012; Grimey Drawer

I feel like we're getting into the extreme worst-case scenario for someone who needs to back up their data. I mean, if you're lugging around a library of congress-sized library, then yes maybe you should take a little time to make backups. Otherwise the cloud is a reasonable alternative to backups.

what the hell were we talking about again?

# ¿ Jan 19, 2017 21:13

Setset: Apr 14, 2012; Grimey Drawer

I seem to have stumbled upon an old Mac Pro with a Xeon W3520 processor in it. Are these things good for anything? Google says roughly equiv to a i7 920

# ¿ Mar 10, 2017 19:59

Setset: Apr 14, 2012; Grimey Drawer

well poo poo, threw it on craigslist for $150 and sold within the hour o.O

anyway, that's my story. thanks for listening

# ¿ Mar 11, 2017 02:07

Setset: Apr 14, 2012; Grimey Drawer

Otakufag posted:

When can we expect budget mobos to go with a i5 8400?

intel's release says 1st half of 2018... so that probably means you have to wait until May

# ¿ Oct 7, 2017 23:03

Setset: Apr 14, 2012; Grimey Drawer

underage at the vape shop posted:

Helping my brother build his pc, where's a good place to read about motherboards. Probably going with the i5 8400, and on the cheaper end.

Intel seems to have rushed this launch. The only motherboards available are the Z series which are not necessary for an 8400 chip. Next year there should be the B/H model boards that will be more suited to the 8400 processors, at a lower price.

# ¿ Oct 10, 2017 11:51

Setset: Apr 14, 2012; Grimey Drawer

Struensee posted:

It's cute that you think it's unintentional

Pretty sure I never expressed my opinion one way or the other, but thanks for reading so deeply into my post

# ¿ Oct 10, 2017 18:19

Setset: Apr 14, 2012; Grimey Drawer

eames posted:

The coffee lake paper launch was really quite genius, Intel sent a bunch of 5.2 GHz binned samples to the youtubers, a handful of retail boxes to e-tailers and now gets to watch Ryzen sales fall off a cliff until Q2/2018.

Doesn't it hurt Kaby lake sales too, though? Oh well, I guess Intel can take a sales hit as long as it affects AMD too. Boo

# ¿ Oct 10, 2017 23:22

Setset: Apr 14, 2012; Grimey Drawer

Has there been an IPC comparison with the last few generations yet? IIRC Skylake/Kaby lake had almost 0 IPC gain at all. I think they have to move to 10nm for actual gains, but I've been wrong before

# ¿ Oct 13, 2017 12:09

Setset: Apr 14, 2012; Grimey Drawer

So, the thing about water cooling is: it's great at absorbing heat and can absorb a lot of it. But when it gets to a certain point, the radiator won't be able to dissipate it fast enough, which will lead to normalizing temperatures that are somewhat similar to air cooling.

My advice is to not stress test your half-decade old CPU for long a duration

# ¿ Oct 14, 2017 23:17

Setset: Apr 14, 2012; Grimey Drawer

Paul MaudDib posted:

Icelake is a gamble. Die shrinks often hurt the overclock you get (see: Ivy Bridge, Broadwell). Intel has obviously been strugglebussing with this node, do you trust that just because Intel says the node is 10nm+ that they actually hit the same clocks as 14nm++?

Coffee Lake is on a mature node, it's a huge step in multi-threaded performance, it's the best single-threaded performance on the market. It's obviously the next Sandy Bridge as soon as it's available in quantity. Ice Lake is going to be delayed/have clock problems/yield problems/some other bullshit. As soon as this poo poo is at Microcenter with decent mobos for $120 and processors for $330 you should snap it up.

Downside: a lot less PCIe lanes and a lot less RAM bandwidth, in an era where NVMe and fast networking are both becoming affordable

Those who have moved on to 10nm (pretty much just Apple via TSMC) are having decent IPC gains, 20-30%. You're making a lot of negative assumptions about Ice Lake based on little evidence. The real question is, how does AMD respond to Coffee Lake? and will it put pressure on Intel to release Ice Lake early like they did with Coffee Lake?

# ¿ Oct 15, 2017 12:51

Setset: Apr 14, 2012; Grimey Drawer

Would the 8700 be a good choice for streaming? or does that still require a second pc with a capture card?

# ¿ Nov 23, 2017 13:48

Setset: Apr 14, 2012; Grimey Drawer

Does disabling the iGPU in BIOS not give the same thermal headroom?

# ¿ Dec 29, 2018 15:15

Setset: Apr 14, 2012; Grimey Drawer

Khorne posted:

The timescale of the attack seems impractical for most valuable consumer data.

As the vulnerability becomes more widespread it could easily be worth a hackers time to go after consumers. Anything that can be automated will be and spending a few processor seconds on it isn�t a big deal

# ¿ Mar 7, 2019 18:44

Setset: Apr 14, 2012; Grimey Drawer

14nm +IX

apple and amd have already paved the roman numeral future for everyone

# ¿ Mar 27, 2019 22:42

Setset: Apr 14, 2012; Grimey Drawer

Agreed posted:

Got my 4770k to 4.4GHz at 1.28V right now, input voltage of 1.9V, and it seems quite stable. Using an Asus Z87 Sabertooth board, it has been straightforward enough to overclock.

Should I be concerned at that voltage though? I read 1.25V is considered safe on air, and I guess I am a bit past it, but I haven't had bad thermals except under Prime95 which is such an outlier for heat. Under gaming conditions and general use I've not seen higher than ~60�C.

Kinda weird waiting to overclock until 4 years after I put together the comp (with then-2 year old parts) I know but it just kept being enough until suddenly it wasn't, you know? I still don't run into many day to day computing scenarios that kill it, but with a RTX 2070 from EVGA's 700/900 step-up promotion on the way, I bet I will be seeing its limits in some modern games. Should at least be a bit better at 4400ghz versus the 3500ghz stock, I guess.

If you set the voltage to 1.9 and are only seeing 1.28 I would be worried that the reading isn�t accurate. Do you know about LLC and how it works?

# ¿ Jul 3, 2019 17:12

Setset: Apr 14, 2012; Grimey Drawer

Paul MaudDib posted:

it is completely bonkers how everyone has gone all-in on 3D stacking without any idea how to cool it, everyone�s just sure it�s the next big thing.

DRAM on die (or die on dram) makes a certain amount of sense but nobody has working solutions to cool big powerful dies in the middle of the stack

The solution could be as simple as "dont run this at 5ghz". IPC should go up since everything is on the same substrate, so the question is will it go up enough to make 2ghz chips match performance of their high clocked counterparts

# ¿ Oct 12, 2019 16:05

Setset: Apr 14, 2012; Grimey Drawer

Indiana_Krom posted:

Bringing DRAM closer to the die is for bringing data closer to the CPU so it can wait less and work more, which means it will also consume more power. Stacking everything together will definitely make cooling and power consumption worse in pretty much every way, because it sticks more power consuming and thermal dissipating components in less space than ever before and helps them all work harder besides.

I hope you don't think that they plan to make these chips 95w TDP. 7nm doesn't scale well with high power usage. They will try to target the optimal spot of the efficiency-curve. More likely we'll get similar performance of today's chips but at 15w or 7w so it can be passively cooled and tossed in a tablet or surface pro

# ¿ Oct 12, 2019 16:35

Setset: Apr 14, 2012; Grimey Drawer

PCjr sidecar posted:

Bruh, AMD is shipping a 280W 7nm part today.

yeah a 5700xt pulling 280w gets like 3 fps more than one using stock 180w. hence the poor scaling

# ¿ Oct 12, 2019 18:00

Adbot: ADBOT LOVES YOU

# ¿ May 13, 2024 15:42

Setset: Apr 14, 2012; Grimey Drawer

PCjr sidecar posted:

Yeah that�s what I�m referring to. 30% higher base clock for 40% higher TDP relative to 200w part. 7nm may or may not scale well but its not going to stop vendors from putting as much in as possible.

Original comment was referring to 3D stacked dies using 7nm. There would be no way to cool a 280w part (well, eight 35w parts) in a situation like that. So I was simply stating that Intel would probably try to hit peak efficiency curve rather than max out the clock speeds.

But yeah look at the efficiency on those 8-core chiplets on EPYC. From 65-95w tdp @ 3.6-w/4.4ghz boost to what, 25w at 2-w/3.3ghz turbo?

# ¿ Oct 12, 2019 22:19

The Something Awful Forums > Discussion > Serious Hardware/Software Crap > Intel: lol