**UPDATED: 3DMARK RESULTS** Intel Core i9-9900K, Core i7-9700K and Core i5-9600K CPU Performance Leaks *+Release date rumour*

[IceCold008]

Still debating if it worth upgrading my 4770k to the 9900k or wait until next lineup.  And im not talking about money thats not the issue. In terms of preformance its the "Do i need or Do I want" question. Its more of a want than a need i think

[porina]

22 minutes ago, suits said:

I'm worried that the reason they soldered ther chip is because of thermals and lack of overclocking headroom. It's a very "Intel" thing to do. The 8700k and 7700k have had pretty high thermals with 4 and 6 cores, hopefully they didn't move to soldered to make 8 cores just "do-able". 

The paste was deemed adequate by Intel for up to 6 core parts. Adequate of course does not mean best. For stock, it'll do. Even for light overclocking, it isn't the end. I managed to get 5.1 all cores bench stable from memory on my 8086k without delid, and on air cooling.

 

It is likely that for the 8 cores, even at stock Intel can't achieve their performance targets without going to solder. That'll lower CPU temps and buy more headroom. I wouldn't expect any worse overclocks than current. In that sense, I kinda agree that Intel isn't doing what it would do otherwise with more cores, that is to drop the clocks somewhat. They want to maintain the clocks. Not surprising, as they will want to ensure performance domination over AMD.

 

Ryzens don't exactly have a lot of overclocking headroom either. With upgraded air cooler my 2600 will turbo all cores to 3.9. Manual overclock got me to 4.2 and climbing the voltage wall. Maybe I can push a bit more but the voltage is going to be scary. The 8086k stock is all core turbo to 4.3, and as said most if not all are getting 5.0+ without difficulty.

[JoeCoke]

5 hours ago, PianoPlayer88Key said:

I often hear people getting excited about a tiny 5-10% or even 25% performance jump year over year, or maybe 50% over 5 years or so.

 

Am I the only one here who, although it was in my childhood and I"m having to look up benchmarks after the fact, remembers performance jumps of like 50 to 100 *TIMES* over a previous system?

 

 

Of course everyone wants some huge amazing jump in IPC, but it's not likely to happen any time soon, it's just not realistic. Of course I know about the times where we'd get amazing jumps in performance, but those times aren't comparable to now. We've hit a massive law of diminishing returns. To see something that dramatic, I think we'd have to move on from x86. Or who knows, with AMD being competitive, Intel might suddenly drop some bomb within the next few years, like when they dropped the Core 2 CPUs after years of having half the IPC of the Athlon 64.

[Void Master]

2 minutes ago, 2Buck said:

5 hours ago, PianoPlayer88Key said:

I often hear people getting excited about a tiny 5-10% or even 25% performance jump year over year, or maybe 50% over 5 years or so.

 

Am I the only one here who, although it was in my childhood and I"m having to look up benchmarks after the fact, remembers performance jumps of like 50 to 100 *TIMES* over a previous system?

 

 

Of course everyone wants some huge amazing jump in IPC, but it's not likely to happen any time soon, it's just not realistic. Of course I know about the times where we'd get amazing jumps in performance, but those times aren't comparable to now. We've hit a massive law of diminishing returns. To see something that dramatic, I think we'd have to move on from x86. Or who knows, with AMD being competitive, Intel might suddenly drop some bomb within the next few years, like when they dropped the Core 2 CPUs after years of having half the IPC of the Athlon 64.

Who knows let's wait and see.

Maybe even VIA could dominate the market later .

who knows what the future holds.

[The Benjamins]

5 hours ago, PianoPlayer88Key said:

WORDS

 

I often hear people getting excited about a tiny 5-10% or even 25% performance jump year over year, or maybe 50% over 5 years or so.

 

Am I the only one here who, although it was in my childhood and I"m having to look up benchmarks after the fact, remembers performance jumps of like 50 to 100 *TIMES* over a previous system?

 

 

 

Oh ... btw a Tom's Hardware article from several years ago (Intel's 15 most unforgettable x86 CPUs) says the 286 was about 3.6 times faster than the 8086 at the same frequency.

 

If all x86-compatible CPU generations since then had the same performance-per-clock-per-thread uplift, how fast should an Intel Core 9000 series CPU be?

 

 

 

It is called diminishing return. We are not going to get any more major improvements out of clocks or IPC on silicon, we are nearing the limits of physics.

 

It is like why are cars 0-60 times not having the same improvements year over year like it did in the past.

Their is only so much you can do with the current materials used.

[JoeCoke]

Just now, Void Master said:

Maybe even VIA could dominate the market later .

Ha, wouldn't that be beautiful? All these years of Intel VS AMD and VIA (who pretty much NO ONE besides us geeks knows about) comes "back from the dead" and dominates everything.

[Void Master]

Just now, 2Buck said:

Ha, wouldn't that be beautiful? All these years of Intel VS AMD and VIA (who pretty much NO ONE besides us geeks knows about) comes "back from the dead" and dominates everything

Realy, and AMD vs NVIDIA for years and then Intel GPUs dominate the market !!

That would be sooooooo exiting,

[DrMacintosh]

So exactly why should I consider one of these rumored chips over an R5 2600? I’m not impressed by these numbers. 

[DrMacintosh]

1 hour ago, IceCold008 said:

Still debating if it worth upgrading my 4770k to the 9900k or wait until next lineup.  And im not talking about money thats not the issue. In terms of preformance its the "Do i need or Do I want" question. Its more of a want than a need i think

You don’t. 4770k vs anything modern would provide little performance gain in real world use unless you are encoding video or something heavily multithreaded. 

[Aetheria]

24 minutes ago, Void Master said:

Who knows let's wait and see.

Maybe even VIA could dominate the market later .

who knows what the future holds.

I can foresee a future in which Intel moves on to other markets, AMD gets sick of losing money and caves, and grumpy old me tries to relive the glory days of the x86, only to find that VIA have 2W embedded chips with 8700k performance.

[Void Master]

1 minute ago, Aetheria said:

28 minutes ago, Void Master said:

Who knows let's wait and see.

Maybe even VIA could dominate the market later .

who knows what the future holds.

I can foresee a future in which Intel moves on to other markets, AMD gets sick of losing money and caves, and grumpy old me tries to relive the glory days of the x86, only to find that VIA have 2W embedded chips with 8700k performance.

I am waiting for 35w chips with i9 7980XE performance !! 

Good luck VIA !

 

[GoldenLag]

28 minutes ago, DrMacintosh said:

So exactly why should I consider one of these rumored chips over an R5 2600? I’m not impressed by these numbers. 

no good reason. on the lower end of the new CPUs there is no change. maybe a bit boost clocks, but otherwise the same. 

 

id be excited for the i5 9400 if it wasnt the exact same part as the 8400. 

 

another thing is are we getting 400 series mobos from intel?

do we have to dish out to the Z390 to enshure mobo support?

if not, will we have a program to borrow CPUs to update the BIOS?

 

it just seems like Intel:

1. upped the cost

2. moved up a tier in the product stack and removed ht at i7 to justify the i9. 

 

 

good news is. we now have another current gen 8 core 16 thread CPU. its only 200$ more expencive when accounting for the cooler. 

[Tech Enthusiast]

3 hours ago, schwellmo92 said:

So we're supposed to trust your "thinking" over actual statistics?

1

No, I made it abundantly clear that is my personal thinking and that I don't know much about CPUs. I don't expect anyone to "trust" me on this. It is just what I think about it, nothing more, nothing less.

3 hours ago, schwellmo92 said:

The Intel equivalent might be 20% faster in non heavily-threaded workloads, but at more than 20% cost, and then in heavily threaded workloads it loses, so you would be

paying more for a worse processor in certain scenarios.

3

Paying 20% more for 20% more performance is pretty great. Usually, it is more like 50% more for 10% more performance.

3 hours ago, schwellmo92 said:

And then looking at gaming benchmarks on their own is also inaccurate, how often is the only thing you have running on your PC the game? For me and everyone I know, almost never. It's not as clear cat as you are making it sound.

So, you don't like benchmarks. Fine, I get that and I see it plenty of times. But how do you pick a CPU then? Trust in other people that claim to know more? Trust in marketing? Trust in companies? Brand loyalty? Pure Price? If every Benchmark that does not look like you want it to look, you can hardly use them, right?

I know Benchmarks are not the full picture, but they are a 1:1 comparison between CPUs. If one is doing it better, it is likely faster, no?

What makes you think that one CPU is doing better with Game + other stuff than another CPU doing the same? Do you guess? On which basis?

 

So many questions, but I hope you get my point. If nothing a review tells us is true because we believe it has to be false,... than reviews need to change in order to be helpful again.

[Brooksie359]

10 minutes ago, Tech Enthusiast said:

No, I made it abundantly clear that is my personal thinking and that I don't know much about CPUs. I don't expect anyone to "trust" me on this. It is just what I think about it, nothing more, nothing less.

Paying 20% more for 20% more performance is pretty great. Usually, it is more like 50% more for 10% more performance.

So, you don't like benchmarks. Fine, I get that and I see it plenty of times. But how do you pick a CPU then? Trust in other people that claim to know more? Trust in marketing? Trust in companies? Brand loyalty? Pure Price? If every Benchmark that does not look like you want it to look, you can hardly use them, right?

I know Benchmarks are not the full picture, but they are a 1:1 comparison between CPUs. If one is doing it better, it is likely faster, no?

What makes you think that one CPU is doing better with Game + other stuff than another CPU doing the same? Do you guess? On which basis?

 

So many questions, but I hope you get my point. If nothing a review tells us is true because we believe it has to be false,... than reviews need to change in order to be helpful again.

Most people use benchmarks to measure performance of cpus but most people like to use reliable ones. Geekbench has been know to be quite unreliable and doesn't give a good picture of real life performance. 

[MageTank]

10 hours ago, porina said:

Assuming the other stuff you wrote is correct (as in, I haven't bothered to check) then all it means is only look at Geekbench for non-HT performance. Not everything benefits from HT.

 

Cinebench (assuming R15) may be well known and predictable, but it is also one possible use case. We know it scales well with cores, gets a nice boost with HT/SMT, and isn't very ram sensitive. Still, not representative of all use cases.

 

On average, an 8 core no HT part would probably perform better most of the time. 8 cores are 8 cores. 6c12t could be equivalent to 6 to 9 cores, depending on the workload. Cinebench R15 would be equivalent to around 8 cores and it would be uncommon to see much more gain than that. 

We really can't use Geekbench as a credible metric of performance at all for PC. It's impossible to know how fast a CPU or memory kit is because Geekbench only reports DMI strings, not actual clock values. You can have a 6ghz CPU and Geekbench will still only report it's base/boost clocks in the result. It also only reports JEDEC ram speeds even if you were to load an XMP that goes beyond JEDEC standards. 

 

Geekbench is only useful for confirming specifications. Every piece of performance information should be disregarded until actual performance based benchmarks are used.

[RobbinM]

OK guys the first 9700K full review just arrived:

https://elchapuzasinformatico.com/2018/09/intel-core-i7-9700k-review/

[Guest]

20 hours ago, GoldenLag said:

Just fyi that has veeeeeeery litttle to do with core-to-core latency. Its has way more to do SSD speed and max clockspeed.

 

Even then the difference between them in day to day tasks is as far as i know unnoticable.  

 

If there is a core-to-core latency affected workload then there will be a crippling of performance (also bandwidth has a major role aswell, see the 2990wx) though very few workloads are affected. 

 

Yes the core-to-core latency advantage intel holds is undeniable due to the ringbus, but that advantage sadly gets less and less due how the ringbus works. Eventually the meshbus on the skylake X is better. From what i can gather the effects of this latency is not as you describe it

My W3565 on an mSATA SSD through a shitty adapter felt snappier than my R5 1600 at 3.8GHz on an 850 EVO.

[GoldenLag]

3 hours ago, Leinad4400 said:

My W3565 on an mSATA SSD through a shitty adapter felt snappier than my R5 1600 at 3.8GHz on an 850 EVO.

Well...... Either you are experiencing placebo or there is something affecting your system. They should be more or less the same.

[mrthuvi]

12 hours ago, MageTank said:

Geekbench is only useful for confirming specifications. Every piece of performance information should be disregarded until actual performance based benchmarks are used.

If there's even a one percent chance that the benchmark is trustworthy, we have to take it as an absolute certainty. 

- This sub every time there's a rumor confirming their bias. 

[Brooksie359]

14 hours ago, RobbinM said:

OK guys the first 9700K full review just arrived:

https://elchapuzasinformatico.com/2018/09/intel-core-i7-9700k-review/

What are they running those benchmarks at? 1440p? Because it looks like all those games are pretty gpu bottlenecked. 

[Guest]

9 hours ago, GoldenLag said:

Well...... Either you are experiencing placebo or there is something affecting your system. They should be more or less the same.

Same applies for a G4560 + Sandisk SSDPLUS rig that I still have. Have you tested both side by side yourself? It is possible that it's something with my system, but I've heard the same complaint from many people.

[GoldenLag]

1 hour ago, Leinad4400 said:

Same applies for a G4560 + Sandisk SSDPLUS rig that I still have. Have you tested both side by side yourself? It is possible that it's something with my system, but I've heard the same complaint from many people.

Interesting. I havent noticed anything between the pentium system i have (probably 2 gen old pentium, cant check atm) and my Ryzen system, but i havent had a side by side comparison. 

 

If this is a thing, what would cause it? Because cache latency and core-to-core latency wont. 

[RobbinM]

New 3DMark results added!!! Have fun guys.

[JuNex03]

18 hours ago, mrthuvi said:

If there's even a one percent chance that the benchmark is trustworthy, we have to take it as an absolute certainty. 

- This sub every time there's a rumor confirming their bias. 

I knew batman was running this sub

[TahoeDust]

5 hours ago, RobbinM said:

New 3DMark results added!!! Have fun guys.

Intel Core i9-9900K Flagship 8 Core CPU Tested in 3DMark Timespy Benchmark – Faster Than An Overclocked Ryzen 2700X at Stock Clocks

Yesterday, we got to see the first media review of the Core i7-9700K and a day before that, we saw many Geekbench 4 benchmarks of the entire unlocked 9th Generation processor lineup. The CPUs are still some time away from their official launch and many reviewers and overclockers already have their hands on the processors and Z390 motherboards. Some of the earlier benchmarks we got to see were posted by motherboard manufacturers themselves who forgot to pull off the plug from the internet where the results got listed (accidentally).

 

The performance results show that at stock clocks, the Core i9-9900K 10,916 points and 36.68 FPS in the CPU tests on 3DMark Timespy. The chip was also tested by a different user with clock speeds of 4.8 GHz across all cores and that scored 11,459 points and 38.50 FPS in the CPU tests. Now when we compare these results to an overclocked AMD Ryzen 2700X at 4.45 GHz that seems to be the near limit of the AMD flagship 8 core processor, the chip scores 10,285 points and 34.56 FPS in the same CPU benchmark.

 

 

 

 

Interesting.  Slower than my 7820x at the same 4.8GHZ...the cpu score by 6%.

 

Link to comparison: https://www.3dmark.com/compare/spy/4374217/spy/2216374#