Intel’s Core-X i9 and i7 series (x299) & Xeon (1P/2P)

Damn I thought all were Clubbed together mobile, LGA and Xeons too.

 

No... this is an old thread, within days of being a year old about the new release of x299 CPU's and motherboard's back then...

Dr. AMK accidentally posted a bunch of info here instead of in the new thread, and all those posts were deleted or moved here:

Intel Core i7-8750H/ i7-8850H/ i9-8950H Coffee Lake
http://forum.notebookreview.com/thr...0h-i9-8950h-coffee-lake.810891/#post-10706961

 

Maybe @Vasudev was blinded by the i9 branding

 

Yeah. I thought they folded everything into one product instead of umbrella products with different features and silicon quality and bins.

 

No, they aren't at all related, are produced in different production runs, in different locations, and using different processes.

Wafers aren't cut up and used across products, they aren't made to be able to fit into different carriers for different sockets, that's fantasy thinking overrunning reality.

Let's stop here now, and take the new product discussions into the right threads, all of this is OT here.

 

TOTAL Ripoff from Intel?
Linus Tech Tips
Published on Apr 23, 2018

 

Intel’s 9th Gen CPUs Need A New Motherboard AGAIN?!

 

Intel Announces EOL Plan for Kaby Lake-X Processors
by Anton Shilov on May 1, 2018 12:00 PM EST
https://www.anandtech.com/show/12701/intel-announces-eol-plan-for-kaby-lakex-processors

"Intel this week revealed plans to discontinue its Kaby Lake-X processors. The chips will not be supported by the company’s upcoming X399 platform for high-end desktops, so initialization of their EOL program is not surprising. Interested parties will be able to get their Core i5/Core i7 processors in LGA2066 packaging for about a year, but they will need to order the chips by the end of November.

Intel on Monday announced plans to discontinue all versions (tray and boxed) of its Core i5-7640X and Core i7-7740X CPUs. PC makers and component resellers interested in these processors will have to order them by November 30, 2018. Intel will ship the final codenamed Kaby Lake-X chips by May 31, 2019, so technically interested parties have a year to buy these chips if they need them.

Intel introduced its Core i5-7640X and Core i7-7740X CPUs in mid-2017 in order to enable hardcore enthusiasts and professional overclockers to set overclocking records using quad-core Kaby Lake-X CPUs while taking advantage of the company’s latest HEDT platform. Usage of the X299-based motherboards with LGA2066 form-factor ensures better power supply to processors and thus helps to hit higher clocks. The plan was heavily criticized by product reviewers and motherboard makers since Kaby Lake-X CPUs require different voltages and memory kits than the high core count Skaylake-X CPUs. Furthermore, the launch of Intel’s six-core Coffee Lake processors in October made Kaby Lake-X products considerably less attractive."

Intel Admits It Made A MAJOR Mistake!
UFD Tech
Published on May 2, 2018

 

https://www.hardocp.com/article/2018/04/09/intel_rumors_kaby_lakex_skylakex_cascade_lake

"Intel is looking to target bins pushing ~5GHz on 12 and 14 core parts (WOW!) with an estimated TDP range of 275w to 300w on Socket R. This TDP bump will assuredly will require new VRM layouts on motherboards, so we will see some new boards out to address..."

@Mr. Fox @TANWare @ajc9988 @tilleroftheearth @Papusan time to go full desktop and ditch laptop.

thanks to AMD upping the core and zen's IPC, intel realized the only real way for them to stay ahead before giving out 10nm+ would be to push more cores as well as having frequency lead. this might be a sign of saying good bye to IPC improvement and just frequency push from this point on, P3 all over again.

in the end, time will tell, intel got a lot of fixes before i'd jump in. few things to improve includes:
- overclocking mesh frequency 4000mhz +
- heat/efficiency (comes with 10nm+ and beyond)
- turbo boost max 3.0 that actually works

 

Intel will give away thousands of 8th Gen Intel Core i7-8086K Limited Edition processors to celebrate the 40th anniversary of its x86 architecture.
https://www-techrepublic-com.cdn.am...es-how-to-enter-to-win-one-for-your-business/

 

That's not an x299 i9 CPU, it's a z370 i7 CPU...again you posted in this thread with OT content...

Here's a good place to discuss that special edition CPU:

Alleged Intel Core i7-8086K 40th Anniversary Processor Leaked and Benchmarked – 6c/12t , 5.1 ghz !
http://forum.notebookreview.com/threads/alleged-intel-core-i7-8086k-40th-anniversary-processor-leaked-and-benchmarked-–-6c-12t-5-1-ghz.815636/page-4#post-10740528

 

X299 is soon dead. This is what’s coming...

I really hope AMD this time can offer a lot more than 16 cores for next (coming) Ryzen Threadripper... Intel Unveils 28-core/56-thread HEDT Processor-Techpowerup.com

Intel crushes Computex with a 28-core chip, a 5GHz 8086K, two new architectures and more-Pcworld.com

As we all know... Last time was AMD trumped by 18 core i9-7980Xe. Will we see the same happen this time as well?

Edit.
Intel Shows Off 28 Core / 56 Thread HEDT X299 Processor, Destroys Every Other HEDT Chip – Also Promises Mainstream Platform CPU Updates in 2018

 

The arrival of Intel's 28-core CPU at 2:47

 

Time to start another Intel CPU / Motherboard chipset thread, as "x299 is dead..." - @Papusan

How about one of you two do it this time? I'm swearing off / at Intel until the new architecture CPU's arrive next 18-24 months.

 

Sure... Intel’s Flagship Coffee Lake-S 8 Core CPUs Arriving in September on Z390 Refresh Platform – Skylake-X Here To Stay, Will Get 22 Core CPU Update This Year.
http://forum.notebookreview.com/thr...99-xeon-vs-epyc.805695/page-224#post-10744594

Valid info for old/current chipset/MB

 

HW News - Threadripper 2 TDP, Intel Kills Rumors, & X399 Board
https://www.gamersnexus.net/industry/3341-hw-news-threadripper-2-tdp-intel-kills-rumors

 

Intel i9-9980XE Review: Disappointing Overclocker, But Good Stock
Gamers Nexus
Published on Nov 13, 2018
The Intel i9-9980XE didn't perform exactly as we expected. Our review of the CPU looks at stock and overclocking performance, including extensive thermals. Article: https://www.gamersnexus.net/hwreviews...
This review benchmarks the Intel i9-9980XE vs. AMD Threadripper 2990WX, Intel i9-7980XE, 9900K, and more. Testing includes a focus on overclocking and thermals, with additional testing for power consumption, Premiere rendering, Handbrake transcoding, Photoshop, gaming tests, and more.


Intel i9- 9980XE Review - ANOTHER Skylake Refresh?
HardwareCanucks
Published on Nov 13, 2018
The Intel i9-9980XE is an expensive processor that's supposed to perform well in benchmarks against the AMD Threadripper 2 2950X. But at $2000 its VERY hard to recommend.

 

Intel Announces Cascade Lake Advanced Performance and Xeon E-2100
Cascade Lake advanced performance represents a new class of Intel Xeon Scalable processors designed for the most demanding high-performance computing (HPC), artificial intelligence (AI) and infrastructure-as-a-service (IaaS) workloads. The processor incorporates a performance optimized multi-chip package to deliver up to 48 cores per CPU and 12 DDR4 memory channels per socket. Intel shared initial details of the processor in advance of the Supercomputing 2018 conference to provide further insight to the company's extended innovations in workload types.

Cascade Lake advanced performance processors are expected to continue Intel's focus on offering workload-optimized performance leadership by delivering both core CPU performance gains1 and leadership in memory bandwidth constrained workloads. Performance estimations include:

• Linpack up to 1.21x versus Intel Xeon Scalable 8180 processor and 3.4x versus AMD EPYC 7601
• Stream Triad up to 1.83x versus Intel Scalable 8180 processor and 1.3x versus AMD EPYC 7601

Intel Shares More 48-Core Cascade Lake-AP Xeon Performance Data Vs AMD EPYC
Processor core counts are skyrocketing these days, both on the consumer side and of course in the workstation sector. In regards to the latter, Intel last week announced the immediate availability of its Xeon E-2100 processors targeted at small and medium size businesses, and also a 48-core Cascade Lake CPU for burlier workloads. According to Intel's own testing, the 48-core chip trounces AMD's Epyc 7601 processor.

Let's back up a moment. Intel is gearing up to launch a line of Cascade Lake-SP Xeon processors built on a 14-nanometer plus-plus (14nm++) manufacturing process, set to debut before the end of the year. Intel also announced that it will plans to launch even higher performing Cascade Lake-AP processors during the first half of 2019, and they will be powerhouses compared to the current Xeon Scalable family, as well as AMD's Epyc 7601.

That's where the 48-core chip comes into play. It beats AMD's 32-core/64-thread Epyc 7601 in core and thread counts, while we'll have to wait and see how it stacks up in terms of clockspeeds and overall cache. In the meantime, Intel claims its upcoming slice of silicon tops AMD's monstrous part with a 3.4x uplift in LInpack and a 1.3x uplift in Steam Triad. And compared to the Xeon Scalable 8180, the performance delta is 1.21x and 1.83x, respectively.

Intel is now providing some more performance metrics based on a set of real-world applications, rather than synthetic tests. As shown in the slide above, Intel's performance claim compared the Epyc 7601 amounts to the following:

• MILC: Up to 1.5x
• WRF: Up to 1.6x
• OpenFOAM: Up to 1.6x
• NAMD (APOA1): Up to 2.1x
• YASK (ISO3DFD): Up to 3.1x

It's fair to say that the average user is not familiar with any of those applications. In short, they represent a variety of high-performance computing (HPC) modeling and simulation tasks, from quantum chromodynamics to weather and atmosphere research and prediction. Heavy duty stuff, in other words, and of interest to the scientific community.

Of course, with Intel's 48-core Cascade Lake-AP having more cores and threads than AMD's Epyc 7601, we would naturally expect better performance in applications that can take full or at least partial advantage of the additional processing resources. So, this isn't as much about the architecture as Intel may want to portray, as it is about core and thread counts. It's also worth noting the fine print, and specifically that SMT is on with two threads per core in YASK and NAMD, and one thread per core in some others, while still other tests have SMT disabled, which in some cases could be a function of the test itself.

Nevertheless, these are impressive performance claims, if they hold true. The performance advantage could also be short lived—AMD recently showed off a 64-core/128-thread Epyc "Rome" processor based on its 7nm Zen 2 architecture.

 

It's all-around bad when even a notorious Intel boot licker is throwing shade your way.

 

Have you not seen the Zen2 Epyc coverage? Or that the 64 core is likely in the same power envelope as the 32 core (so 180W-250W), whereas expected power envelope of the AP professor is minimum of 350W estimated from what I've seen, as well as them turning off SMT on the AMD servers, although Intel argues it is proper to do for linpack to give a higher score and they have effectively done so in other ways through affinity settings in prior test beds.

I'm actually less impressed with this chip overall, especially since a prototype 64 core chip can own a server with 2x28 core chips, which is more than a single 48-core chip.

Then factor in the improved floating point performance, which improves AVX instructions, the changes to IF2 and the I/O chip getting rid of NUMA on socket while this adds NUMA on Intel's side, plus questions of how they are using the UPI interconnect, I'm REALLY unimpressed with Intel's offerings, including the 9th series HEDT that dropped this morning.

Sent from my SM-G900P using Tapatalk

 

As usual, I will wait to form an opinion after we see how well it overclocks. This applies to literally everything... AMD, Intel and NVIDIA products. Stock performance doesn't impress me unless it is pretty decent to start with and there is tons of extra performance to be gained by overclocking the crap out of it. This is how I decide on good versus great, and this is the primary reason nothing presented by AMD holds any interest for me thus far. I am waiting for that to change, and hopefully it will some day. And, as usual, I am not interested in speculation or theoretical discussions. I know you enjoy that, which is totally fine. That's just not my thing and I choose to take the wait and see approach before choosing a winner.

 

Then go watch GN saying 7980XE is better.

Sent from my SM-G900P using Tapatalk

 

GN did a video showing exactly how well (or not in this case) the new HEDT CPU overclocks. It's linked on the previous page.

 

Exactly. What it allows is they used 14nm+, like I called, soldered the chip, and called it a day. The AVX512 multiplier is the same, AVX multiplier is higher by 100mhz, and all core boost is higher by 400MHz at stock. LM provides more cooling than the solder, which we already knew from the 9900K, so everything matches my prior prediction, which many ignored.

I don't GAF if people don't want to speculate, but when they divorce themselves from reality to maintain their view...

As I said, no upgrade needed unless you are a professional that doesn't OC the processor. Otherwise, underneath is practically same chip as the 7980XE. People thought the 14nm++ would be used. Nope. So at best, better binned from a Kaby lake process line due to 14nm+ maturity.

Cascade-X gets 14nm++, but will primarily just be Xeons. Good job, Intel!

Sent from my SM-G900P using Tapatalk

 

Thank you. Yes, I saw it and I watched it. I'm not replacing my 7960X any time soon. I'm not done benching the crap out of it yet. I won't replace it until I've extracted everything I can from it and and I am no longer able to consistently earn points with it at HWBOT, or doing so becomes an overly difficult hit or miss endeavor in terms of stability in striving for a higher overclock. When it's time for that I will look at all the available options and choose the one that seems like the best option for overclocking. If AMD has the best option for extreme overclocking performance, that's what I will buy. If not, it will be another Intel CPU.

 

I doubt AMD TR or Rome can OC higher since they aren't manufactured to break OC WR but give consistent CPU performance for demanding tasks such as Rendering, CPU based ML modelling etc..
I feel you can tweak XFR to hit upto 4.5GHz if you use your portable AC setup. Its just a theory.

 

Maybe so, but hard to say. Using the wrong tool for a job often doesn't end well.

 

Yes you are right.

 

Intel Core i9-9980XE vs AMD Ryzen Threadripper
Kyle Bennett, Tuesday , November 13, 2018
https://www.hardocp.com/article/2018/11/13/intel_core_i99980xe_vs_amd_ryzen_threadripper/1

"Today Intel is kicking off its newest High End Desktop processor, the Intel Core i9-9980XE Extreme Edition. This 14nm Skylake-X CPU boasts 18 Cores and 36 Threads and has an expected retail price of $1979. We compare the i9-9980XE to AMD's entire line of Threadripper CPUs to see where the 9980XE sits in the HEDT stack.
...


The 9980XE seems to overall fall in between the 2950X and 2970WX performance-wise. From that perspective, you are paying about $679 to $1079 more for the 9980XE, than the competitive Threadripper processor. The one outlier here being instances of H.264 encode on the WX parts. However even looking at those two H.264 examples, you might be better off purchasing a 2950X which is well over $1000 less than the 9980XE.

The Core i9-990XE in in no way a "bad" CPU for content creators, or for gamers for that matter. The 9980XE simply seems to be overpriced for the performance value it brings to the table now that AMD and its Threadripper are on the scene in the HEDT market. From an enthusiast perspective, Intel has not left many clocks on the table unless you have top end cooling and then you are still going to likely be pushing the 9980XE to uncomfortable temperature levels."

 

Yeah, I agree that it's way too expensive. We're living in some pretty bad days overall when you stop and think about the sad state of affairs in PC technology right now. Pretty sad.
And, for the record, I also think the Threadripper CPUs are severely overpriced. For the price they are asking, they are not a good value either. They suck at overclocking, and for the price they are asking they should be really good at overclocking. I guess you have to pick your poison and save a few hundred bucks, or step up to the plate and pay out the nose for overclocking. Either option really sucks. I'm thinking we should all stick with old tech and take a vacation from PC tech for a year or two until Micro$lop and all of the hardware OEMs/ODMs can gets their acts together and come back with good stuff we can all appreciate. When you stop and think about all of this nonsense, there is almost nothing that is actually worth being happy about right now.

Edit: OK, one silver lining I forgot about... RAM and SSD prices are finally falling and getting closer to representing good value, so we can upgrade those and wait on everything else, LOL.

 

Wth are you talking about? My first gen TR, depending on task, is 25-30% slower than your 5GHz chip while running 1GHz slower and costing $700 less. The second gen TR 2950X costs $900 vs the new chips or 7th gen chips costing way more. For those not using chillers, doing open loop like I am (which I am heat, not voltage, limited), the difference is even smaller.

Now, I'll agree, if you can wait, wait for 7nm third gen TR or Ryzen, or Epyc 2. Same with cascade (Intel's 14nm++ line) or Cooper (14nm++ refresh). I know you don't like speculation, but here goes. The reason Intel didn't deliver 14nm++ chips for HEDT is they won't have 10nm (equivalent to TSMC 7nm) until 2020 for server and HEDT. That (14nm++) is potentially up to 10% more performance they could release next year. Considering what AMD has shaping up, saving it to go head to head with TR3 makes a lot more sense (especially since manufacturing woes will be less by then). That won't have much IPC gains, but will likely have the 6 channel memory for HEDT and higher clocks, which should give good uplift. But, if being honest, I don't think it will be enough for the Zen 2 changes.

Either way, Microsoft and software vendors need to get off their butts because both AMD and Intel chips are starting to break the OS, grow beyond its scheduler, etc.

Sent from my SM-G900P using Tapatalk

 

This is what I am complaining about. I don't want/like that. I'd rather pay more and get more, and that 1GHz and 25-30% speed gap is why I reject AMD as an option. I am unwilling to compromise on that to save ~$700. So as much as I hate getting my wallet raped, it is what it is. Same reason I won't settle for something less than an X080 Ti GPU. I could save lots of money on a Vega, X060, X070 or X080, but I am not interested in PCs as a hobby any more if it involves settling for second-fiddle hardware. If playing games was my hobby any of those products would be good enough.

 

You are the cause for that pricing, so you have no right to complain. This is capitalism. When you show you will pay unreasonable prices for relatively small increases in performance, you tell those companies the market can bear more. So saying you don't like it now because it is getting too expensive is you becoming aware of your break point. Yet you promote higher on mainstream, which is driving that price point up as well. You as well as I know if the sales, though smaller, give the same or more profit, but you produce fewer goods, then the new floor is set. Only something disruptive, like charging half that amount for 80% of the performance, can break the price floor after breaking peoples conceptions, like your own.

But, unlike the CPU side, the GPU side doesn't have that force, meaning of enough people are dumb enough to get the 2080 Ti at a 70% price premium for 30% more performance (Intel's 20% gains on the 9900K for 70% more cost is the same thing), then the floor gets set on pricing and without a disruption, that is the new normal. Yet consumers didn't reject either, meaning welcome to new pricing! I'm hoping AMD keeps the price pressure on next year.

Now, there are few people that need the chips for work, including press, content creators for workstations, pro overclockers, and some hobbyists. But too damn many think they are in a group they are not. All this led to very binned chips and pay to play for top benching, destroying the hobby in part. This isn't to say don't participate, rather be aware.

Meanwhile, even the few times AMD has ever had the lead, people used then for price pressure on Intel and Nvidia instead of getting the best chips. Next year, 7nm, at 25% performance gain, would have performance parity on servers and mainstream, and close to that on HEDT (remember that gas in the tank of 10% or so from 14nm++, it's why AMD would need even more than that to not just win on price to performance in the segment, but straight up performance).

But for most, buying the second best chip FOR HALF the price (which isn't just a "few hundred dollars") is fully worth it.

Not just that, we are also seeing changes in board design, we are setting plans to utilize pcie 4.0 and 5.0 for storage, and AMD outright said server gets DDR5 in 2020 (if same I/O chip is done on TR4, we could see them do DDR4/5 compatibility like Intel did on some boards/chips for DDR3/4 (or just make chips backwards socket compatible, but new socket can't use old chips or something like that)). That means they are trying to force Intel to use DDR5 for ice lake in 2020, which I/O and IMC shrinks are more difficult, especially with a process with huge issues already.

Sent from my SM-G900P using Tapatalk

 

Also, here is Intel's cascade bench numbers (remember, cascade AP has 1.5X the cores of Epyc, and Epyc 2 will have 1.33x the cores of this chip and many uarch changes):
LinpackNumerical Linear Algebra 3.4x
Stream Triad Memory Bandwidth 1.3x
MILC Quantum Chromodynamics 1.5x
WRF Weather Forcasting 1.6x
OpenFOAMComputational Fluid Dynamics 1.6x
NAMD (APOA1)Molecular Dynamics 2.1x
YASK (ISO 3DFD)HPC Kernel Tuning 3.1x
https://www.anandtech.com/show/13586/intel-offers-more-cascade-lakeap-performance-numbers

Sent from my SM-G900P using Tapatalk

 

Fair enough, but everything I said that you responded to is still true. Capitalism is not perfect (nothing is) but it's the only model that actually works. The problem isn't only price. It's a combination of price and mediocrity, which is even more damning. And, it's why I said, " I'm thinking we should all stick with old tech and take a vacation from PC tech for a year or two until Micro$lop and all of the hardware OEMs/ODMs can gets their acts together and come back with good stuff we can all appreciate." Part of the problem for AMD (besides their current offerings suck at overclocking) is credibility. They are the root cause of that. They spent so many years in the shadows and selling garbage that it is still really, really difficult for me to embrace anything from them yet. The fact that their GPUs still suck and how I got repeatedly burned by their broken trash makes it even more difficult to let bygones be bygones. Not saying I won't eventually get over it, but they're not bringing enough (for me) to bury the hatchet yet. They're getting lots closer with the CPUs. Now all they need to do is take the next step and beat Intel at overclocking. I'm watching... waiting.

I am not optimistic they will ever do enough to earn forgiveness and deserve a place at the same table as NVIDIA. And, I am not giving them any wiggle room for nonsense. Being the value leader doesn't impress me and that is just maintaining status quo for AMD. I don't give credit for sucking less than they used to.

But, the good news for them is that I am actually watching. That alone is a first for them in about a decade and borders on miraculous. I can appreciate the progress. And, to be fair, Intel's recent blunders and missteps do not represent more than a decade of wallowing in abject failure.

 

Don't let Apple hear you, they will see iMac Pro or a modified Cylinder case server that throttles every time.
I don't think AMD can OC higher like Intel can in the next 3-4years and at the same time Intel can't compete with AMD at price/perf ratio in the next 5 years until they release new architecture in smaller nm lithography.

 

Why you SHOULD NOT delid your i9-9980XE

 

Although capitalism being the only model that works is empirically false, not the place to discuss that point.

As to AMD, you mentioned being burned before. Everything should be on any company earning a consumer's dollar. This is why no one should be pre-ordering and everyone should be doing their own analysis. If you look at the carve outs I put in the value argument, high ends hobbyists, like those with chillers for example, are mentioned. But, I do disagree with how you frame higher clocks to mean higher performance. To less tech savvy people visiting the forums, they may think frequency is king when it is frequency x IPC that tells the story, and IPC varies by task, meaning each CPU could have its wins and losses depending.

Now, Intel, with soldering, is leaving less in the tank, just like AMD. Nothing wrong with that, but you need to stop using that talking point as well.

So we finally get to overall performance. Pure and simple, this is all that matters. Not frequency, not IPC as a stand alone number, but the mix of it. This is the reason to buy any product. AMD must still earn that dollar, regardless. It shouldn't be a wait and see a response, or wait until a preferred company catches up. If they can earn it, then vote with that cash. I have this same problem with electoral politics. People will vote for their "team" even when the person they are voting for doesn't represent ANY of their values, policy positions, etc. Don't vote on a wish and a fart. If a party had shown themselves to lie, to tell voters one thing and do another, etc. (Applies for all current political parties depending on politician and varying by degrees, like vote blue no matter who crap or voting in a right wing nut (not all right wing are nuts)). This is why we vote power person, or in the economy by per product.

Here, AMD is theoretically going to tie Intel next year, or maybe slide a bit ahead. The following year looks similar, even with Ice Lake vs Zen 3. Right now, I've for no reason to think AMD will win on frequency, but every reason to think they will win on IPC. But until reviews are in, they haven't earned my money.

It's like Asus Z390 MBs. They usually make such good products, but there are others that provided more for less as regards power delivery. It isn't saying all Asus is bad, it's saying this gen sucks and they are not worth it this gen. Can it do the job? Sure. But other boards can do it, do it better, and at a lower price point.

TL;DR - companies must earn peoples money. Too many fall for the false argument of not buying until the company they like (brand loyalty - extreme form is fanboy; but not all brand loyalists are fanboys) puts out a good product or using competition just to lower their price on their brand. That is anti-competitive and it kills competition creating monopolies and the pricing that goes with it. Anyone saying they want AMD to be competitive so they can get a cheaper Nvidia GPU don't believe in competitive markets. Same on Intel side if using the argument (was seen before Zen release more often). Buy the best product at the moment. That is how you force support by software vendors through changes in market share.

Sent from my SM-G900P using Tapatalk

 

Frequency means nothing. If IPC is high enough, a 4.5GHz CPU can outperform an over 5GHz chip, which that is expected.

For example, let's say Intel had 25% more performance at 5GHz vs AMD at 4.1GHz. Let's add 10% to frequency, giving AMD 4.5GHz all core, then give 13% IPC improvement. 1.1*1.13=1.24, meaning 24% more performance, which would put those two CPUs in the same performance category, within a percent. This shows that overclocking isn't just about frequency. It is about going over spec to get more performance. If all you want is frequency, go play with the AMD FX series. Lol.

So, with AMD claiming in official press release a 29% IPC in mixed floating point and integer workload, while a leaked IPC of 13% in scientific workload came out, both being early silicon and not knowing which major or minor revision was used (AMD was on the F major revision for Zen and minor revision 4 at CES before the release), we could see something like the above hypothetical in practice.

For HEDT, Intel kept 14nm++ held back in case, so AMD would need to exceed that for that platform, whereas on mainstream and server, we are already seeing AMD do well against 14nm+ for released numbers. So no reason to frame the argument that frequency is absolutely needed for top OC performance.

Sent from my SM-G900P using Tapatalk

 

Pardon me but I fail to understand your point.
I agree with you about IPC. But, simply relying on IPC alone won't win any games/benchmark since most developers have/use Intel Compilers that provide greater analytics/insights on how to optimize their code on multiple Intel CPUs and use general or un-optimized code paths for competitors. If AMD can succeed in that Compiler part where even the general code runs optimally w/o actually needing specific workarounds for AMD TR/Ryzen/FX separately. Hardware wise AMD seems to have picked up.
Then again, at the end of the day I'd take more cores over higher single thread IPCs because in long term High core counts really matters for Deep Learning,Neural network etc..
EDIT: Mr. Fox simply likes higher GHz and wants to top every benchmark charts.

 

You are correct, you miss my point. I'm saying highest GHz does NOT equal top score in all cases. That is why I pointed to the frequency beast of the AMD FX chips.

You have a point on compiler optimizations, although Intel knows now forcing the worst unoptimized path if not Intel chips is illegal and costs them lots, and that people are looking for it. Also, Intel is adopting two NUMA nodes on single socket, so will face the same issues on jumping die to get to the memory controller as AMD saw on Epyc and TR with their Cascade AP chips.

But, as said, if the instructions per cycle is higher on AMD, then the frequency, or number of cycles in a period of time, can be lower to accomplish the same performance.

Sent from my SM-G900P using Tapatalk

 

Actually, after thinking about it with fresh eyes, you are wrong. IPC is the measure of instructions per cycle with a specific task. That means, aside from using like GCC or Clang optimized compilers, you already see the impact in performance of using non-optimized compilers in the final IPC number since it tells a lot on how the program is to be used by the hardware. So if the gain in IPC is there, that is post any sandbags from the compiler issue.

Sent from my SM-G900P using Tapatalk

 

Don't forget MSFT does have a compiler and you can choose to optimise specific architecture or generalize it.
Most compilers are aware of ARM,Intel,AMD etc.. and fine tuning will take many years at software level. 2-3 years ahead IPC of AMD and Intel will more likely similar if Intel still use 14nm.

 

But, the point I made in the last post is that the IPC already bakes in the performance at a specific task (which is why a range of IPC is more appropriate way to discuss IPC, but I digress). This means that, with time, the IPC can be improved through adoption of different compilers or compiler updates, as well as software vendors learning more optimizations. But, if an unoptimized compiler is currently used, that is baked in before the IPC numbers we look at, so isn't as much of a concern if discussing IPC as it currently stands, instead just being another area in which IPC can be improved.

 

Learning How to RIP Der8auer with Liquid Nitrogen, ft. Steponz
Gamers Nexus
Published on Jan 27, 2019
Learning how to overclock using liquid nitrogen, featuring a 9980XE, dual Titan RTXs, and Bearded Hardware.
The stream is at 1PM Eastern Time (1 hour from publication) on the 27th. It will be streamed on YouTube.
Find Joe's channel here: https://www.youtube.com/channel/UCHcG...
And his Instagram here: https://www.instagram.com/steponz1/
Walking through liquid nitrogen overclocking with "Overclocking Celebrity" Joe Stepongzi, we learn the basics of how to use liquid nitrogen to achieve higher overclocks on an Intel i9-9980XE. Two Titan RTXs in NVLink are used under chilled water, as previously, with the EVGA X299 DARK as the platform.

 

Live: Liquid Nitrogen Overclocking a 9980XE, ft. Joe Stepongzi
Gamers Nexus
Streamed live on Jan 27, 2019
This is our first liquid nitrogen overclocking livestream! We'll walk through the LN2 XOC process, get some basic numbers in, and prepare for future OC battles by learning how to LN2 OC!
Find Joe's channel here: https://www.youtube.com/channel/UCHcG...

 

Stream Recap: First Time Liquid Nitrogen Overclocking
Gamers Nexus
Published on Jan 28, 2019
Recapping our recent liquid nitrogen overclocking livestream with Joe Stepongzi. We hit about 5.8GHz on the 9980XE and need to work on actually doing SLI testing next.
Find Joe's channel here: https://www.youtube.com/channel/UCHcG...

 

This may need to be moved to the Sky-X thread, but, 9990XE spotted in the wild: Puget Systems Lists Intel Core i9-9990XE Specs, Explores Cooling

New: Premiere Pro CC 2019: Intel Core i9 9990XE Performance

"If 9900K & 9940X had a love child, 9990XE would be it"

Highlights: 14core, 4GHz/5.1Boost, 255w TDP, auctioned to OEMs only, "Puget Systems will have benchmarks of the Intel Core i9-9990XE up in the next week or two"

 

Intel X299-refresh, VRM Temp Test: MSI X299 Creation, Asus X299-Deluxe II & Gigabyte X299 Aorus
Hardware Unboxed
Published on Mar 5, 2019

 

Deliding Intel LGA 2066 CPU WITH A VICE !!!
Timmy Joe PC Tech
Published on Mar 13, 2019
Do you need a delid tool or can you use common household tools to delid your Intel CPU? Let's check before and after using liquid metal!