Intel 8700k 11% faster than 7700k

[TVwazhere]

6 minutes ago, goodtofufriday said:

ryzen 2 what you got for me

THATS AN EVEN NUMBER

WE CANT HAVE EVEN NUMBERS IN NAMING SCHEMES. THATS LIKE WANTING AN ODD NUMBER OF CORES

DO YOU WANT REALITY TO IMPLODE ON ITSELF?

[TVwazhere]

1 minute ago, Shiv78 said:

they probably will at some point once i7s are 8C/16T and i5s are 6C/12T 

Potentially, but by then AMD with have 16C/32T main stream CPU's

[pas008]

5 minutes ago, TVwazhere said:

Was* 40% less, now it's more like 20% ($80-90[Used to be 65-70] vs $110) Intel finally realized what shit having a pentium that good was

AFAIK, i3's have never had multi threading, and most likely never will. Definitely not 4c/8t, who'd buy an i5 then?

https://ark.intel.com/products/97527/Intel-Core-i3-7350K-Processor-4M-Cache-4_20-GHz

[Bananasplit_00]

47 minutes ago, MyName13 said:

Holy sheet, imagine if they just released a new bios that unlocks cores, maybe that's why all ryzens have identical power consumption

Anyway this might make ryzen obsolete, especially r3 and r5.

No they don't... How did you get that in your head? 

 

43 minutes ago, Dietrichw said:

Hasn’t intel claimed this kind of IPC improvement before but when tested in real world cases it isn’t close to the improvement that they claimed.

Yep, that's basically standard operation it seems. 

[NumLock21]

6 minutes ago, Sniperfox47 said:

In past leaks the way we've seen them list it is:

Base clock/6-core boost/4-core boost.

 

This puts it about on par with the boost of the i7-7700k

Do we know what that's based on? Iirc Coffee Lake has a revised GPU alongside it right? It wouldn't be the first time Intel has included iGPU performance in benchmarks of their CPU improvement.

Based on that slide picture. Did you even look at it?

[Bananasplit_00]

10 minutes ago, TVwazhere said:

THATS AN EVEN NUMBER

WE CANT HAVE EVEN NUMBERS IN NAMING SCHEMES. THATS LIKE WANTING AN ODD NUMBER OF CORES

DO YOU WANT REALITY TO IMPLODE ON ITSELF?

I WANT ODD NUMBERS OF CORES THOUGH, TRICORE DREAM LIVES ON!!! 

[TVwazhere]

Just now, Bananasplit_00 said:

I WANT ODD NUMBERS OF CORES THOUGH, TRICORE DREAM LIVES ON!!! 

HEATHEN!

[porina]

16 minutes ago, Taf the Ghost said:

Considering "multi-thread" is 51% gain, it looks more like the 8700k simply turbos harder than the baseline 7700k. And probably a little bit of efficiency in its turbo. I made a joke about Speed Shift v3 being unannounced, but that might actually be true. There's room to pick up single-threaded performance in other spots beyond the main core. Maybe the memory system can feed a single core better.

Simply moving from 4 to 6 cores at the same clock is 50% peak improvement. 51% seems awfully specific, so if it some kind of measure that is based on, there may be something else going on.

14 minutes ago, done12many2 said:

Skylake-X needed to go to mesh to include the 14 - 18c parts as well.  That by no means indicates that this will happen in mainstream.  I guess we'll see, but what you are saying is that because they went to 6 cores on mainstream, they are probably going to mesh, when Intel has been able to do just fine up to 10 cores in the past without doing so. 

Past rumours did list claimed L3 cache sizes, and based on those, Coffee Lake I think is more similar to ring than mesh models. Personally I'm really hoping for ring structure, as extracting the potential performance out of my 7800X I think has been hindered by mesh. Software basically has been optimised heavily for ring for so long, it isn't performing as well as it could.

9 minutes ago, TVwazhere said:

AFAIK, i3's have never had multi threading, and most likely never will. Definitely not 4c/8t, who'd buy an i5 then?

i3s have HT. At least my Haswell and Skylake ones do, they're all 2c4t parts.

 

In my testing HT gains you a maximum 50% more potential CPU throughput than not having it. Thus a 4c8t part AT BEST will equal a similarly configured 6 core part. For non-HT friendly code, you essentially have a 4c4t part. In other words, for most things, a 6c6t part will be faster than a similar 4c8t part at same clock.

[Bananasplit_00]

Just now, TVwazhere said:

HEATHEN!

YOU KNOW NOT YET THE POWER OF THE TRICORE(may or may not be because there has been like a total of one ever released commercially back on FM2+) THE GLORY, THE PERFORMANCE, THE COMPLETE-MENTAL-SCREWUP! THE THREE HOLY PARTS ALL IN ONE! 

[Sniperfox47]

10 minutes ago, NumLock21 said:

Based on that slide picture. Did you even look at it?

...yes... I see it... I can't read Chinese so I have no idea if it says where they got that 11% or how it was benched...

[Herman Mcpootis]

7 minutes ago, Bananasplit_00 said:

I WANT ODD NUMBERS OF CORES THOUGH, TRICORE DREAM LIVES ON!!! 

your dream's stuck in 2008 m8, might be time to move on 

[done12many2]

2 minutes ago, porina said:

Past rumours did list claimed L3 cache sizes, and based on those, Coffee Lake I think is more similar to ring than mesh models. Personally I'm really hoping for ring structure, as extracting the potential performance out of my 7800X I think has been hindered by mesh. Software basically has been optimised heavily for ring for so long, it isn't performing as well as it could.

 

I'm on the exact same sheet of music as you.  

[porina]

35 minutes ago, Ganz said:

As for price increase I'm torn, part of me says they will increase, part of me says they won't.  I really would believe they would price it competitively with the 1800x or ~$450

Don't forget there's a 7800X already available in 6 core space. I think 8700k has to come in below that, and I hope as position successor to 7700k.

27 minutes ago, Taf the Ghost said:

Maybe the memory system can feed a single core better.

 

I think when we get detailed testing, the single-core will be like 1-3% faster, but if the test was memory sensitive, it'd show more. 

Forgot to reply to this bit... Elsewhere in this forum some of us had tried to find cases where a single thread was ram limited, and we failed. It just doesn't generate enough load to saturate it. Well, maybe if you deliberately ran single channel at low clock you could do something, but not in the real world. So I doubt it will be from the ram side. Cache changes remain a maybe, but I hope they stick to traditional ring as that seems to work better for consumer level workloads.

[Juanitology]

1 hour ago, porina said:

I was too lazy to post this, and there is less here than what I wrote in another thread

 

7700k boost of 4.5, add 11% = 5 GHz. Could Intel be angling trying to sell the first 5 GHz consumer CPU? OR any CPU for that matter... this assumes no IPC change and performance is due only to clock. 51% more multithread could simply be from same clocks but more cores.

 

Similar exercise could be performed for lower models in time.

I wonder if that's 11% higher IPC(which would actually be somewhat of an improvement) or just 11% higher clocks.

[Bananasplit_00]

5 minutes ago, herman mcpootis said:

your dream's stuck in 2008 m8, might be time to move on 

Yah well il just develop my own chip, just got to get hold of a "few" billion dollars first. Chip design is what I want to work with though, super cool stuff. One day you will see a thread on here about the launch of a tri core chip, at which point its probably as useless as the current only tri core lol

[NumLock21]

12 minutes ago, Sniperfox47 said:

...yes... I see it... I can't read Chinese so I have no idea if it says where they got that 11% or how it was benched...

Who cares where they got it from, 11% is better than 0%

[Herman Mcpootis]

17 minutes ago, Sniperfox47 said:

...yes... I see it... I can't read Chinese so I have no idea if it says where they got that 11% or how it was benched...

pic doesn't say anything about where they got the 11% performance boost from, just says generational improvements or smth like that at the top-left hand corner.

[Sniperfox47]

2 minutes ago, NumLock21 said:

Who cares where they got it from, 11% is better than 0%

*Facepalm* let's all believe the unsubstantiated source with no references?

 

And again, it wouldn't be the first time where Intel used a benchmark that included GPU performance in their Single Core performance deltas, so it's possible that those boosts may come from there rather than the CPU itself.

 

Just now, herman mcpootis said:

pic doesn't say anything about where they got the 11% performance boost from, just says generational improvements or smth like that at the top-right hand corner.

Okay thanks. I'll be interested to see the numbers when they announce it over here.

[GatioH]

1 hour ago, porina said:

I was too lazy to post this, and there is less here than what I wrote in another thread

 

7700k boost of 4.5, add 11% = 5 GHz. Could Intel be angling trying to sell the first 5 GHz consumer CPU? OR any CPU for that matter... this assumes no IPC change and performance is due only to clock. 51% more multithread could simply be from same clocks but more cores.

 

Similar exercise could be performed for lower models in time.

AMD has a space heater that boosts to 5GHZ

[porina]

14 minutes ago, Terodius said:

I wonder if that's 11% higher IPC(which would actually be somewhat of an improvement) or just 11% higher clocks.

I severely doubt it is from IPC, unless it is in a specific area where they are adding new instructions to boost it.

3 minutes ago, GatioH said:

AMD has a space heater that boosts to 5GHZ

Without OC?

[GatioH]

2 minutes ago, porina said:

I severely doubt it is from IPC, unless it is in a specific area where they are adding new instructions to boost it.

Without OC?

Yep. FX 9590

[Dylanc1500]

43 minutes ago, done12many2 said:

 

As I said, there is no indicator at this point to even assume that this will apply to the Coffee Lake architecture.  Skylake-X needed to go to mesh to include the 14 - 18c parts as well.  That by no means indicates that this will happen in mainstream.  I guess we'll see, but what you are saying is that because they went to 6 cores on mainstream, they are probably going to mesh, when Intel has been able to do just fine up to 10 cores in the past without doing so.

The biggest reason for the mesh is, the main uses for those chips, scalability. Consumers don't need scalability up to 8 chips and 224 cores on one board... Well... Most...

[Sauron]

1 hour ago, Merp83 said:

If this is true, it will disrupt ryzen - however i Have a feeling ryzen has a dirty update to give more performance...

it will merely catch up.

[LukaH]

different but same. 

[Cookybiscuit]

Please be soldered please be soldered please be soldered please be soldered please be soldered please be soldered.