More Leaked Intel Coffee Lake Benchmarks Appear

[Guest]

More Intel Coffee Lake benchmarks spotted, this time for SiSoftware. Where there's smoke, there's often fire.

More Leaked Intel Coffee Lake Benchmarks Appear : Read more

 

[SteelCity1981]

it's going to be interesting how amd responds with ryzen 2.

 

[TJ Hooker]

Maybe I'm just dumb, but I find SiSoftware's benchmark database impossible to navigate. Can anyone tell me how the leaked results from the article compare to the current high-end, mainstream offerings from Intel and AMD? I.e. 6 & 8 core Ryzen, i7-7700(K).

 

[InvalidError]

it's going to be interesting how amd responds with ryzen 2.

Before worrying about AMD's response, I'd have to see Intel's pricing. I wouldn't be surprised if the 6C6T i5-8600k or whatever it ends up being called ended up costing ~$300, putting it out of reach for most people looking for a ~$200 CPU like the R5-1600.

 

[ffleader1]

RIP Ryzen.

This CPU is going to compete with the 7800X, and that thing costs around $400.
So if Intel prices this too low, they will cannibalize their product.
$350 is a reasonable price, $300 is low but possible, and they both still way way worse price/performance than even the R7 1700.
Also...Nice try replying with a clone account made on July 24. Seriously, you can be an Intel fanboy and have dignity at the same time.,,but you aren't.

 

[kinggremlin]

it's going to be interesting how amd responds with ryzen 2.

There will be a 14nm+ refresh for Zen released next year. Ryzen 2, utilizing a 7nm process, is not expected to see market until 2019. It will not be a response to coffee lake which is expected next month. By the time 2019 rolls around, Intel will have released coffee lake, cannonlake and likely icelake.

 

[Virtual_Singularity]

Nice article, but still to early to rely on for sure (a point that was more or less made in the article). For all we know the info provided may prove false for a variety of reasons. Really still comes down to having to wait & see.

"RIP Ryzen. -6"

Way to early to make such a statement.

 

[10tacle]

I'm due for chipset upgrade on my three year old i5 Haswell. Ready to make the six or eight core leap now that I'm getting into video editing software and seeing some games responding to more cores (that's your crystal ball). I'm holding out for Coffee Lake but if they pull another minimalist improvement with horrid thermal performance, I may very well go back to AMD for the first time in over a decade.

 

[DOORWAY]

You can see my comment

 

[ern88]

Wonder if it will be compatible with the Z170 chipsets?

 

[blockhead78]

TBH... if they have improved IPC and competitive pricing, coffee lake could prove to be very tempting

 

[InvalidError]

TBH... if they have improved IPC and competitive pricing, coffee lake could prove to be very tempting

If there are IPC improvements, they'll be in the usual single-digit range. I'm pretty sure single-threaded IPC potential is practically tapped-out. We're seeing tiny incremental gains simply because attempting to gain more IPC isn't cost/power-efficient.

 

[AgentLozen]

If there are IPC improvements, they'll be in the usual single-digit range. I'm pretty sure single-threaded IPC potential is practically tapped-out. We're seeing tiny incremental gains simply because attempting to gain more IPC isn't cost/power-efficient.

Along the lines with what Invaliderror was saying, Coffee Lake was 4/5 finished by the time Ryzen launched a few months ago. It's probably built under Intel's "+3% IPC, +3% clock speed" philosophy. I don't think we'll see profound changes from Intel for another generation or two.

With that said, I am excited to see a proper review from Tomshardware when Coffee Lake comes out in a few months.

 

[blockhead78]

yeah, I'm not expecting huge improvements as far as IPC goes, but if it's priced right and has decent OC headroom... then i could be very tempted

I've been keeping a keen eye on Ryzen and it'll be interesting to see how coffee lake performs.

I have an upgrade itch that needs to be scratched LOL

 

[JamesSneed]

Maybe I'm just dumb, but I find SiSoftware's benchmark database impossible to navigate. Can anyone tell me the leaked results from the article compare to the current high-end, mainstream offerings from Intel and AMD? I.e. 6 & 8 core Ryzen, i7-7700(K).

It really doesn't matter as this will be an engineering sample if it isn't faked. The clocks are way to low for this to be the real product that will come out. My advice just don't worry about any of this pre-release stuff as its almost always wrong, using engineering samples that you cant extrapolate performance from, or just "fake news" to get clicks.

 

[gggplaya]

3.1ghz normal and 4.2ghz turbo boost. I wonder if it'll be able to sustain 4.2ghz with a decent cooler. If not, then the Ryzen 1600 is a way better buy.

 

[InvalidError]

yeah, I'm not expecting huge improvements as far as IPC goes, but if it's priced right and has decent OC headroom...

As 14nm matures and Intel further tighten yield tolerances, I expect their top SKUs to eat most of whatever margins the chips may have until there is almost none left much like Ryzen. If you look at Intel's boost clocks, one could argue that this is already happening as typical OCs are only 200-300MHz above the top SKU's boost clock.

 

[androbourne]

Well what has history shown us with each jump to a new architecture with Intel? We get roughly a 5-15% increase in IPC. (and mostly on the 5% side).

I highly doubt we will be getting anymore then that with this new release.

It most likely wont even be worth upgrading to from a i7 6700k.

While I'm interested in seeing the results. I'm not getting my hopes up on it being "the badass in town".

 

[spdragoo]

Maybe I'm just dumb, but I find SiSoftware's benchmark database impossible to navigate. Can anyone tell me how the leaked results from the article compare to the current high-end, mainstream offerings from Intel and AMD? I.e. 6 & 8 core Ryzen, i7-7700(K).

No, you're not dumb. I had to tell it to filter the aggregate results by brand (AMD), Platform (Desktop PC), & Operational State ("Normal", to exclude overclocked results), then selected from the options "Ryzen 5 1600X" (their top-line 6C/12T CPU).

http://ranker.sisoftware.net/show_device.php?q=c9a598d994d0f0a2dba1c4aa8abf9fae98a898c0e681bc91a086f4c9f9dfb68bba9cf4c9fcdaa29fae88ed88b585a3d0eddd&l=en

The 1st result was for the same benchmark, Processor Arithmatic (http://ranker.sisoftware.net/show_run.php?q=c2ffcfe988e9d4ecddecddeeddfb89b484a2c7a29faf89fac7f7&l=en):
-- much faster rating (3x the GOPS overall)
-- much, much better Dhrystone results (~4x the Coffee Lake's)
-- much better Whetstone results (~2x the Coffee Lake's)
-- their confidence level was much higher ("Valid Result: Medium Deviation" vs. "Unexpected Result: Outlier")
-- better power efficiency (0.09W TDP/thread vs. 0.15W TDP/thread)
-- more efficient use of energy (almost 2x the GOPS/W TDP vs. Coffee Lake)
-- more processing per cycle (~2x the GOPS/GHz vs. Coffee Lake's results)

Of course, this was only 1 of SiSoft's tests, & their site really sucks if you're trying to find all of the results for a particular user/CPU, so I can't tell how it stacks up in all of the categories.

But to be honest, this isn't looking very impressive. When I looked at Intel's existing CPUs to find a 12T version (http://ranker.sisoftware.net/top_device.php?q=c2ffcfe984e9d4e0d7f183be8ea8c1fccdeb83be8ea8d0eddcfa9ffac7f3c1e794a999&l=en), the Broadwell-E i7-6850K 6C/12T CPU was much more in line with a Ryzen 6C/12T CPU (http://ranker.sisoftware.net/show_run.php?q=c2ffcfe988e9d4ecdce9d1e9dff98bb686a0c5a09dad8bf8c5f5&l=en) than this Coffee Lake CPU is ( & apparently the only reason its individual performance was slightly higher than the Ryzen 5 1600X was because of higher clock speeds, based on their performance-per-GHz & performance-per-watt numbers).

 

[InvalidError]

It most likely wont even be worth upgrading to from a i7 6700k.

Pretty sure there are many people with i5-2400 and up who still feel like there is nothing worth upgrading to on Intel's side yet. I've never kept the same primary PC for four years before and now I'm approaching my i5's fifth anniversary, still with no itch to upgrade in sight.

Upgrading to consecutive generations or even every other one hasn't been worth it for most people and companies in a long time, which is one of the key reasons why PC sales have been getting progressively worse for the past decade.

 

[androbourne]

Pretty much my point. Everyone is getting all hyped for something that most likely wont change the game in the industry. I just think its funny.

Don't get me wrong. I get excited for new tech all the time. Just think people go overboard with the Intel vs AMD side of things. They almost always expect the MAX and we get the medium.

 

[FritzEiv]

I'm always a bit torn on these "leak" articles. On the one hand, they're often fake, we don't know the original source or that source's validity, and I feel as if we are just playing into that echo chamber. In this particular case, the only reason we ran this piece was because we found some extra info. And even then, @PaulAlcorn thinks the SiSoftware numbers are nonsensical ("81W TDP part at 3.1GHz base clock?"). The point of the post, however, was that these leaks typically start coming when something is near . . . I'm betting half of you like to get excited about upcoming launches and you like to read about leaks and info from sources, and the other half feels as if it's just playing into the hype train. Thoughts?

 

[10tacle]

I'm betting half of you like to get excited about upcoming launches and you like to read about leaks and info from sources, and the other half feels as if it's just playing into the hype train. Thoughts?

I never get excited about a new Intel tick/tock anymore. Not since Sandy Bridge which was a significant boost over the first generation i-series. Ever since, as many have alluded to, it's just one small incremental increase notch at a time. I'm still running my six year old Sandy Bridge build as a backup gamer for 1080p. The good news is that we can put more money into GPU upgrades every couple of years and skip years of chipset generations. Intel is shooting themselves in the foot with six years of <yawn> tick/tocks not giving anyone a reason to upgrade their earlier generation chipsets.

 

[InvalidError]

Intel is shooting themselves in the foot with six years of <yawn> tick/tocks not giving anyone a reason to upgrade their earlier generation chipsets.

It isn't entirely Intel's fault, there is only so much parallelism that can be extracted from a handful of threads. While AMD still has some catching up to do, both Intel and AMD are near their practical limit on how much of that can be extracted from typical software instruction flows. The only way to gain significantly more performance is more threads and cores but that requires software to catch up where possible, which could take a very long time still based on how multi-core CPUs have been mainstream for over a decade yet most software today is still only lightly threaded.

More hardware is pointless without a meaningful amount of software written to leverage it. Ryzen may hopefully give software developers the motivation to put in that much extra effort but it may still take a while and be limited in scope.

 

[aldaia]

I smell a Troll.

Jensen_3's profile
Member since: 25 July 2017

RIP Ryzen.

Jensen_3 has edited this post 26 July 2017 06:19:23

Ryzen is better. Downvote if you disagree.