Intel Core i5-9600K Stock and 5.1GHz Overclock Benchmarks Leaked

mlee 2500

Honorable
Oct 20, 2014
298
6
10,785
At first this seemed about right for a part that which doesn't have hyperthreading, like its predecessor...but then I investigated and it turns out the old i5 didn't have hyperthreading either...so....wth?

I think customers will come to regret not having hyper-threading on the i7-9700k, and Intel may also regret it if their gamble in disabling it fails to drive enough enthusiasts to the flagship, more expensive i9 processor.

Then again, we're probably years away from games that are optimized for more then 4 to 6 cores or threads anyway, so maybe only people doing serious multi-media work will care.
 

dudmont

Reputable
Feb 23, 2015
1,404
0
5,660
In regards to real cores vs. virtual cores, real cores beat virtual cores every time. Look at I5-8600k vs I7 7700k.
How much of the slight performance drop compared to I5-8600k could be because of spectre/meltdown hardware changes?
 

salgado18

Distinguished
Feb 12, 2007
928
373
19,370
@MLEE 2500, I believe content creators (games, videos, streamers, artists) can use many cores today, so it will probably be a terrible choice for many people. Gamers and regular users won't see much of a difference (many only care for branding anyway), but I think it will not be as popular as the older i5s were.
 

compprob237

Distinguished
Hm, the AMD Ryzen 5 2600 sure put up a good fight. I wonder what the price premium will be to go to the i5-9600K from the $150-$160 AMD chip. Especially considering these early benchmarks suggests its performance is worse than AMD's chip.
 

mlee 2500

Honorable
Oct 20, 2014
298
6
10,785


I hope that's not the case. Not to trivialize those security deficiencies, but in a consumer product which doesn't run on multi-tenant corporate or data-center platforms, the trade off would not be worthwhile at all.
 

matthew_258

Prominent
Jun 10, 2017
17
0
510
people who play games need fast cores, easy to overclock and stable temps...I would not underestimate a future I-5
 

mlee 2500

Honorable
Oct 20, 2014
298
6
10,785


If you take into account that most games are GPU bound, then the i5 is almost certainly going to be the best value.

If on the other hand a CPU tends to last you 6 years and through multiple GPU upgrades, then what's another couple hundred dollars over a half decade relative to the thousands of hours you use it?

In fact I've found that saving money on mid-tier CPU's just means I end up needing or wanting to upgrade it a year or two sooner.
 

s1mon7

Reputable
Oct 3, 2018
96
4
4,635


Sure the physical cores beat them, but being able to service more threads is very beneficial.
Even in terms of pure performance, surprisingly the 6C/6T 8600k (at 50% more cores) performs pretty much in a similar way to the 4C/8T 7700k, while the 9700k at 8C/8T has only a 33% core advantage compared to the 8700k's 6C/12T, making it a much more questionable choice.

While there has been a long-lasting discussion on hyperthreading in games, the quad core i7s are proving right now that it's exactly where the largest advantage lies, as the last few threads are rarely as demanding (due to games being the furthest from offering a symmetrical load across all threads) and just being able to service those threads without needing the full attention of a single-threaded core is perfectly good enough. Basically while having less cores might lower your fps a bit, those extra threads essentially prevent stuttering caused by having threads on hold. In that regard, I believe a 6C/12T CPU would be a better choice than a 8C/8T CPU, as the latter is likely to choke faster, given enough threads that need to be run asap thrown at it (such as in gaming).

Sure, the problem starts when we run software that's able to divide the workload into >4 (for HT quad-cores) super heavy threads capable of loading CPU cores to the max, but with gaming, that's not yet the case - you usually need to service a few main fast threads and a few less demanding ones popping up, including system interrupts, at the same time. In this regard, the hyperthreaded 6C and 8C will add a similar longevity advantage, especially where games are concerned. Even more so in case of the 6C/12T vs 8C/8T, where the hypethreaded hexa-cores are able to service 50% more threads at a time, which might actually turn out better in the end than having a 8C/8T processor. I suspect the 8c/8t will perform similarly at best when fully loaded by heavy threads (with extremely rare non-HT workloads doing better), and be a worse choice in a lot of workloads that rely on a few main threads and are heavy on high-prority light threads at the same time, which is more typical of where everyday usage has been going.

Tl;dr There are going to be workloads that not only are faster on the hyperthreaded units, but workloads that are able to service unevenly demanding threads at the same time (with a few demanding and a few less demanding threads) will benefit from being able to run those extra processes at the same time, even if each of them will be a bit slower. That is very beneficial in perhaps more scenarios than just having 33% extra physical cores.
 
Oct 15, 2018
1
0
10
I do not think Inter will never let their I5 desktop CPU have hyperthreading they're afraid it compete against their I7 and would hurt there I7 CPU sales.
 

PapaCrazy

Distinguished
Dec 28, 2011
311
95
18,890


Agreed... 9600k@5.2 is benching worse than 8600k@4.9 and everyone is caught up on HT. Bigger question is why can't the newer gen chip outrun the older one?
 

Shumok

Honorable
Aug 19, 2013
47
3
10,545


Ohmigosh. My gaming enjoyment with an i5-2400 and entry level R7 240 must be an illusion. The only thing I have there is stable temps. Goodness me!
 

This is a leak of some unknown chinese site were we have no idea how they tested,for all we know they had a windows update running in the background or whatever(overclocking beyond some point has been known to degrade performance for example) ,wait for official reviews for a number of sites before making this kind of statements.
 

kgreer1985

Reputable
Oct 16, 2018
1
0
4,510
2 things..... the z390 is still on its initial bios.. we all know the first iteration is never optimized..... second, 19 clock 2666 ram? couldnt they find anything slower in the ddr4 category? jesus... throw some 14 or even 16 clock in there
 

dudmont

Reputable
Feb 23, 2015
1,404
0
5,660


read my first post, I give a strong hint at the answer ;). The 9th gen chips are the first that have hardware fixes for spectre/meltdown. I can't guarantee it, but I'm quite sure that this is what explains the performance difference(that and the previous reply to your post about who did the test and what methodologies they used).
 

mlee 2500

Honorable
Oct 20, 2014
298
6
10,785


That sounds logical, but my experience provisioning data-center capacity for massively concurrent workloads suggests that there is more to this math. Namely being that if a core or thread runs sufficiently faster then it becomes available for another job, which it may also complete faster and before two slower threads running those jobs at the same time would.

Ironically, in multi-user systems, total slower processing with greater concurrency may be more satisfying to end users, due to the perceptions associated with pending processes versus immediate response.

I don't doubt that there may be some workloads where an older processor with more threads then a newer one will have an advantage (I see this with 128 thread, five year old SPARC T5's), but more often then not I see faster cores compensating for fewer threads.

Lastly I would observe that the logic you put forth formed some of the basis for building ARM server CPU's with more cores then traditional Intel products (that and lower power consumption). For certain applications immediate or concurrent servicing is just more important then absolute speed. Unfortunately for the companies wading their toes into that particular market, AMD has managed to narrow the gap in providing both high thread count AND high frequency (albeit without the reduction in power consumption).
 

dudmont

Reputable
Feb 23, 2015
1,404
0
5,660

I would surmise that in the coming years, AMD will be releasing some much lower power high core server chips. The advantage they'll have is that with better and better turbo algorithms, those chips will become the best of both worlds. Able to accelerate a few cores to very high clock speeds, when needed, yet most of the time being very low power usage for applications that don't demand a few threads run at high speeds.