Intel i9 9900k to hit 5GHZ - Do you have one Rob?!

[Noel 4,501]

3 hours ago, lownslo said:

It will be what it always is... some folks will see the 9900K as a waste of money, while others will see the 9700K as not enough.  In the end folks will buy what they want...

Greg

Yes that's true but says nothing about the question.  It would be nice to know if there are any differences in IPC when HT is off.  It's called I9, versus I7.  Are there any differences in architecture that would make the I9 do better than an I7, all else (ie clockspeed, and HT off) being equal?  Or are they identical beyond HT and binning?

Edited September 4, 2018 by Noel

[Guest]

11 minutes ago, Noel said:

It would be nice to know if there are any differences in IPC when HT is off.

HT OFF for "most" game/sim applications.  HT increases overall processing latency.  Where HT helps is with tasks that can operate relatively independent of one another (not games not simulators) ... these are applications like Web Servers, Video Rendering, 3D rendering (not realtime).  Why HT ON does NOT help simulators is because simulators "threading" targets the loading the terrain and AutoGen which is I/O intensive which will be slow relatively speaking (even with very fast M.2 SSDs) and HT doesn't help I/O since it's in order of magnitudes slower.

I found a good video on HT ON/OFF side by side ... in all but 1 game HT OFF performed better.  Given that HT OFF is more stable and allows for higher OC there is even more of a reason to NOT use HT for games/simulators.

9700K (with no HT option) will probably be the CPU of choice in terms of value for money for most.

The 9900K with HT OFF and the benefit of 4MB more CPU cache should perform a tad better than the 9700K provide same OC.  Either CPU will be a good choice for flight sims and gaming especially if they can hit 5.5-5.8 Ghz with a good water cooling solution.

Cheers, Rob.

[Noel 4,501]

22 minutes ago, Rob Ainscough said:

The 9900K with HT OFF and the benefit of 4MB more CPU cache should perform a tad better than the 9700K provide same OC.

Cheers, Rob.

The HT on or off I already have a clear grip on thanks to SteveW's approach in how to evaluate that.   In using his method I determined my 3930K performed significantly better w/ HT on than off, so I run HT on.  That could well change w/ an 8 core CPU for several reasons I can think of.

What I do not know about is if I9's architecture is fundamentally faster in terms of IPC than I7's for this series of CPUs.  It sounds like you are saying the 4mb additional cache will have some positive effect.  Is there otherwise any differences in architecture from I7 to I9 that would improve IPC beyond the cache effect?

Edited September 4, 2018 by Noel

[lownslo 592]

1 hour ago, Noel said:

Yes that's true but says nothing about the question.  It would be nice to know if there are any differences in IPC when HT is off.  It's called I9, versus I7.  Are there any differences in architecture that would make the I9 do better than an I7, all else (ie clockspeed, and HT off) being equal?  Or are they identical beyond HT and binning?

Anybody who offers you answers to those questions now will only be conjecturing... or busting and NDA.  The point of my previous post is that it won't matter... some folks will see the benefit of a 9900K while others will bet the 9700K is the best choice for the sim.  IPC, clockspeed, HT ON/OFF will just be debating points then.  Point being... choose the one you believe is best and you're willing to buy.

Greg

[Noel 4,501]

1 hour ago, lownslo said:

Anybody who offers you answers to those questions now will only be conjecturing... or busting and NDA.  The point of my previous post is that it won't matter... some folks will see the benefit of a 9900K while others will bet the 9700K is the best choice for the sim.  IPC, clockspeed, HT ON/OFF will just be debating points then.  Point being... choose the one you believe is best and you're willing to buy.

Greg

Yes Greg I understood your point, which had nothing to do with my question.  My question is based on the premise that IPC varies between processor models.   Doesn't it?   I was looking for an answer to the question which can be objectively evaluated at least in terms of an average of several benchmarks that might corroborate theoretical improvements in IPC that Intel was trying to accomplish with this iteration of processors.   I know how to evaluate HT ON/OFF very clearly, so this once again has nothing to do w/ my question.  One can compare the IPC of my 3930K and someone's 8700K, right?  For example, here's one from PassMark looking at single-threaded performance:

3930K@3.7Ghz = 2,239

8700K@3.7Ghz = 2,704

For practical purposes though I'm sure not technically 100% valid we can attribute the 24% increase to improvements in the architecture of each CPU.  IPC increases PLUS clockspeed increases translate roughly into comparative performance expectations for one CPU over another on a given task.   So the question remains, is there anything inherently superior in IPC of I9-9900 over I7-9700 w/ regard to IPC?  I've assumed but never known if Intel's naming one 'I9' meant there implied something fundamentally different between the two.  Is there?  

[Guest]

2 hours ago, Noel said:

What I do not know about is if I9's architecture is fundamentally faster in terms of IPC than I7's for this series of CPUs.

If you believe the leaked 3DMark benchmarks (Time Spy - DX12) ... the 9900K at 3.1Ghz was about 20% faster than the 8700K at 3.6Ghz ... engineering samples.  So if that is "real" then that implies additional architectural changes to the CPU beyond just the extra 2 cores.

As soon as I can get one, I'll post a compare form my 8700K, 7900X, and 9900K - still hoping for Oct 1st retail release (assuming Z390 MB's are available).

For anyone building a system "now", it would be a good idea to wait for these new CPUs and GPUs ... if they do meet their "expected" benchmark numbers, you'll get far more performance for the same money spent.  But you will still not be able to "max everything out" with 1001 add-ons in P3D or XP11 ... so if that's one's expectation, then it will NOT be realized, not now and not ever ... I would fear the day we can max everything out because that would mean continued development of our favorite simulators and add-ons would have come to an end.

Cheers, Rob.

[SteveW 1,792]

The differences are from all CPU versions featuring differences in for example floating point matrix and other things like cache memory and other clever internal stuff like pre computing future branches of execution on sub cores which can differ in strength. All those things combined give rise to differing performance at the same clock.

[Noel 4,501]

2 hours ago, Rob Ainscough said:

But you will still not be able to "max everything out" with 1001 add-ons in P3D or XP1  Cheers, Rob.

This I've always completely understood, hence I live w/in my means as it were w/ that old 3930K and original GTX Titan and prior systems and continue to use P3D almost daily and enjoy it greatly still.   From guestimates derived from this source or that I think I can expect about a 42% increase in total performance for me by going to one of these two new processors and GPUs, which will be a very significant improvement, so I'm really looking forward to it!  In the end i'm also aware the total experience will ultimately be very similar to what I do every day right now, but with a bit more complexity, but not really all that much.   My 3930K seems to want a little more juice these days so I'm thinking it won't be too long before I have to upgrade so perfect timing!

[westman 257]

The Timespy bench, no info of frq used only info of the ES cpu default info 3100mhz and boost 5000, 

my I9 7920X ES report 2900mhz no turbo when run the timespy bench at 5.0ghz, the NDA is not lifted yet only speculations time will tell

probaly overclocked  5,0ghz with all cores enabled

Edited September 5, 2018 by westman

[fghdgdfdfgfgf 111]

15 hours ago, Noel said:

This I've always completely understood, hence I live w/in my means as it were w/ that old 3930K and original GTX Titan and prior systems and continue to use P3D almost daily and enjoy it greatly still.   From guestimates derived from this source or that I think I can expect about a 42% increase in total performance for me by going to one of these two new processors and GPUs, which will be a very significant improvement, so I'm really looking forward to it!  In the end i'm also aware the total experience will ultimately be very similar to what I do every day right now, but with a bit more complexity, but not really all that much.   My 3930K seems to want a little more juice these days so I'm thinking it won't be too long before I have to upgrade so perfect timing!

C'mon man! 

My coffee Lake runs P3d V4.3 with settings that probably require twice the bandwidth you run and can do it at the very minimum 2x the frame rate. Bottom line: That 3930 would produce an unflyable slide show with the settings I run.   .... I get 10% pop in minimum fps just by taking my memory from 3200mhz to 4133 ..... ....6 cores running 5ghz to the 4 cores of that Ivy bridge on ddr3????

What are you waiting for, again?

[Noel 4,501]

11 minutes ago, FunknNasty said:

Bottom line: That 3930 would produce an unflyable slide show with the settings I run.

What are you waiting for, again?

As I stated, I've learned well you have to live within your means, just like you do, and that is why RobA reminds you you will not be running everything maxed out.  I run all of the scenery sliders to hard right, except when in dense large terminals at which point I might put autogen down to Dense as the lowest.   I am most often flying in FTX regional sceneries.  I use ActivesSky out to 100m w/ variable cloud density according to what I'm after.   I would venture to say if two people walked by our two screens they would be hard pressed to see much difference, truly!   I'm in no hurry my friend because I'm real happy w/ the liquid smooth performance I get, near instant texture update rate at 3440x1440.  The biggest impairment I have is actually out of the GTX Titan.   I frequently have to dial back cloud and terrain shadowing do to GPU load.  Once again, it's easy once you fully understand what impacts what in P3D how to live within your means.  I get off on using FSCaptain, aiming for perfect flights, and use the super easy to process Majestic Q400, f/b PMDG NGX & 777.  I don't get off on a wee bit more complexity--it's hardly worth the cost!

[fghdgdfdfgfgf 111]

58 minutes ago, Noel said:

As I stated, I've learned well you have to live within your means, just like you do, and that is why RobA reminds you you will not be running everything maxed out.  I run all of the scenery sliders to hard right, except when in dense large terminals at which point I might put autogen down to Dense as the lowest.   I am most often flying in FTX regional sceneries.  I use ActivesSky out to 100m w/ variable cloud density according to what I'm after.   I would venture to say if two people walked by our two screens they would be hard pressed to see much difference, truly!   I'm in no hurry my friend because I'm real happy w/ the liquid smooth performance I get, near instant texture update rate at 3440x1440.  The biggest impairment I have is actually out of the GTX Titan.   I frequently have to dial back cloud and terrain shadowing do to GPU load.  Once again, it's easy once you fully understand what impacts what in P3D how to live within your means.  I get off on using FSCaptain, aiming for perfect flights, and use the super easy to process Majestic Q400, f/b PMDG NGX & 777.  I don't get off on a wee bit more complexity--it's hardly worth the cost!

You weren't supposed to come back at me with perfectly reasonable logic ......I fully expected you to comeback at me with the I73930 has six cores not four!  -I'm gonna have to fix my spin machine. :-)

[Noel 4,501]

On 9/5/2018 at 7:57 AM, FunknNasty said:

6 cores running 5ghz to the 4 cores of that Ivy bridge on ddr3????

On 9/5/2018 at 9:22 AM, FunknNasty said:

I fully expected you to comeback at me with the I73930 has six cores not four!

I couldn't tell what you were trying to say because the 3930K is not Ivy Bridge, it's Sandy Bridge-E!  Anywho, the dang thing just isn't dying!  Waiting 6y between systems is great nowadays because improvements in thruput have been so relatively paltry.  I am looking forward to an ~ 42% increase in total performance w/ a new I9-9900K and NVidia RTX 2080 Ti, which is really substantial so very happy I have waited this long.

Edited September 9, 2018 by Noel

[fghdgdfdfgfgf 111]

26 minutes ago, Noel said:

I couldn't tell what you were trying to say because the 3930K is not Ivy Bridge, it's Sandy Bridge-E!  Anywho, the dang thing just isn't dying!  Waiting 6y between systems is great nowadays because improvements in thruput have been so relatively paltry.  I am looking forward to an ~ 42% increase in total performance w/ a new I9-9900K and NVidia RTX 2080 Ti, which is really substantial so very happy I have waited this long.

Yeah, as is often the case with me -post 1st research later, that 3930 is a nice processor  ...like an early coffee lake prototype. Anyway, with the new kit I'm gonna bet you a beer that the "42%" benchmark  you have set will be insignificant after you get some seat time with it. You will probably never use the term FPS again.

Edited September 9, 2018 by FunknNasty

[Noel 4,501]

39 minutes ago, FunknNasty said:

 Anyway, with the new kit I'm gonna bet you a beer that the "42%" benchmark  you have set will be insignificant after you get some seat time with it.

Once again, I have no idea why this particular comment.  I'm exceedingly aware of exactly how the sim performs, what impacts the CPU, the GPU, and both combined, and I predict it will be very easy to quantify the difference between the current system and 142% of the current system.