Ryzen vs i7 (Mainstream); Threadripper vs i9 (HEDT); X299 vs X399/TRX40; Xeon vs Epyc

Sure they can. The I/O chip used for Epyc will not be used on mainstream, instead, a second I/O chip will be developed and used. If all is taken in power savings, you can get a decently low powered option to go in laptop devices.

Main question is trying to force GPU tech with it.

So, XMG is a rebrander of Clevo. Clevo is the one making the MB and laptop. If they chose not to use AMD desktop chips because of fear of sales, then that is their decision, not AMD. So it isn't fully in AMD's court.

Edit: Also, they support Asus and I think Acer in their offerings. So contradicted there.

Sent from my SM-G900P using Tapatalk

 

I thought we were talking Ryzen (gen2) Zen2 with @ghegde, not the current AMD mobile offerings. I also believe XMG was referring to Ryzen (Zen2) LGA in a mobile solution (a laptop with Ryzen 2700X not the Ryzen Pro mobile platorm). I also don't consider Asus, Acer and even MSI in any of this... all of their products are of the BGA garbage variety, which is not a device I would recommend to anyone unless purchased for under $600 USD.

In the end, when an ODM develops something using Ryzen/Zen2 (or somehow performs some layout magic with a TR) using LGA in a laptop that can rival the Clevo P870TM1 and an i9-9900x/i9-9700x, then you have something worth talking about.

 

Well, you seemed to miss my point and play favorites, then. XMG sells rebranded clevos, just like Eurocom and other distributors. Since those are Clevo products, what good does AMD supporting distributors do, they would need to work with Clevo. And why are companies that used Zen 1 desktop chips in mobile laptops relevant? Maybe because they did that already, not just mobile CPUs, although with the 50% power reduction, you could have the power of a 2700X in a 48W TDP package, meaning even the mobile chips will get a nice benefit with Zen 2.

So look up which laptops utilize the desktop 8 core chips in a mobile package and try bugging Clevo to create such a product as an alternative to the other implemented products currently on the market.

Sent from my SM-G900P using Tapatalk

 

Not sure about favorites, @hegde and I are not talking Zen / Zen2 BGA mobile platform, but rather a Zen 2 desktop LGA solution - more specifically a full blown 2700X within a laptop. Please do not conflate Ryzen mobile with this sub topic of Ryzen/Zen2 desktop inside a laptop.

In regards to laptop ODMs, Clevo is the only ODM with enough forward thinking to offer any kind of LGA solution at all, so there's no real point at talking to other BGA peddlers. The problem is, Clevo doesn't solicit feedback from the end users, but rather they obtain information through their resellers. When enough ppl gripe to XMG, Sager, HIDevoultion, XoticPC, AVA Direct, Eurocom, etc. this info is making its way back to Clevo. At least that is what is reported by the resellers.

So when the resellers hear back (in this case through XMG) that AMD won't support LGA packages in a laptop, it's a safe guess it will not be happening.

Now, since I neither work for Clevo, a reseller, or AMD, I cannot testify to the entire veracity of what @XMG has shared, but I have no reason to doubt their word either.

 

https://www.acer.com/ac/en/US/content/predator-model/NH.Q3GAA.001
https://www.asus.com/us/Laptops/ROG-Strix-GL702ZC/

That is the desktop 2700, not bga.

Edit:
Sent from my SM-G900P using Tapatalk

 

Excellent! TY!! I didn't see these listed on AMD's site for solutions for Laptop, and was unaware someone actually has come through.

The Helios looks like it is the only one with a 2700. I have some research to do to see if the cooling could handle a 2700X.

Hopefully if units Acer units start to move, we'll start to see other ODM jumping on this bandwagon. Now if they can only get a ThreadRipper to fit in a DTR!

 

Yes, it seems Asus was first gen and I think had a 580 (one reason gamers didn't go for it vs a 1080 in other laptops). Acer came through this year, and although a Vega 56, it is within reason for a laptop DTR.

Now, on TR in a laptop, I'd like to see the range done. So, for the isoperformance, they can use half the power. A 1950X was 180 Watts. That means for the same performance, Zen 2 could possibly do it in a 95W TDP. Or in a desktop, 25% more performance for the same wattage of 180W. So may not be as outlandish as first blush.

Sent from my SM-G900P using Tapatalk

 

Both Ryzen PGA laptops have NBR threads:

Asus ROG GL702ZC owners lounge
http://forum.notebookreview.com/threads/asus-rog-gl702zc-owners-lounge.809882/

Acer Predator (Vega 56+Ryzen 2) Helios 500
http://forum.notebookreview.com/threads/acer-predator-vega-56-ryzen-2-helios-500.817796/

 

Top 5 Best CPUs November 2018, AMD Prices Intel Out of The Game!
Hardware Unboxed
Published on Nov 21, 2018


Awards: Best CPUs of 2018 (Gaming, Production, & Disappointment)
Gamers Nexus
Published on Nov 21, 2018
The best CPUs of 2018 includes a look into gaming, Blender, video editing, and overclocking performance. As a bonus, we also talk about the most disappointing CPU.
Whether you're looking for the best CPU for Photoshop, Premiere & video editing, Blender, or just the best CPU for gaming, we have your 2018 benchmark round-up to show the best. Budget gaming CPUs, high-end CPUs, and workstation processors all make an appearance.
Article: https://www.gamersnexus.net/guides/3398-best-cpus-of-2018-gaming-photoshop-production-streaming

 

the laptop imho is just not worth it. probably because i already have one thats 6 core and can upgrade to 8. for a new purchaser pov its probably a great buy, im not too sure of the cost but its acer + AMD so the value probably very good. theres still BGA GPU people would have to think about tho.

for any AMD laptop, my guess is that some of these OEM will likely put a 16core ryzen in it just like acer did with 2700, try to grab media attention and see how they can sell it for a premium as being the "FIRST" 16 core laptop ever.

even with zen2 chip being very capable, my primary concern remains the same as a laptop enthusiast, TDP. with AMD i have a tad less to worry because their TDP includes turbo boost, while intel's doesn't. these OEM will still design as small of a heatsink as possible to reduce cost and make it thin, and thats biggest down fall for us overclockers.

basically i still have to wait for clevo, wait for unlock bios, wait for zen2 16 cores, wait for some software support (preferably TS).

or we go desktop 32 cores 4.5ghz thanks

 

@ole!!! - I'm not a gamer, so the BGA based GPU bothers me only from a service point of view.

However, I was disappointed when I tried to shop for one that you you couldn't expand / replace items like memory, SSD, and CPU at time of purchase.

 

Don't know if people missed it, but AMD is releasing a speed optimized Epyc chip called the 7371. If they do this on Epyc 2, but include a speed optimized 32 core chip, or 16 core, with PBO active and allowing basic voltage offsets with PBO (including negative voltage offsets), a build like this is very obtainable for those considering TR for HEDT. This got overshadowed by Zen 2, but worth a gander. https://www.anandtech.com/show/13594/amd-launches-highfrequency-epyc-7371-processor



https://www.newegg.com/Product...oduct.aspx?Item=9SIAD7H7D63922
7351P = $1077
https://www.amazon.com/Supermicro-M...41827&sr=8-4&keywords=SP3 motherboard
MB = $461
https://www.newegg.com/Product...duct.aspx?Item=N82E16820147651
8x32GB 2666@CL17 =8*$380=$3040
TOTAL = $4578 for 32 cores of goodness, 256GB of ram goodness, and 128 PCIe lanes. For half that ram, you can get ECC 2666@CL17 for $1664 for 128GB, which isn't bad at all for registered DIMMs, and reduced the cost to $3202 for those components, which really isn't too far off of a workstation build and makes me want to get those memory lanes and PCIe lanes. Sometimes people need to just see what is there for their use cases.

Sent from my SM-G900P using Tapatalk

 

The biggest problem is VRM's of server boards are usually highly lacking.

 

I agree. I'm hoping supermicro or giga do something about that with next gen server boards. SuperO is supermicro's overclocking enthusiast boards which currently only have Intel chipset. With the overclocking Xeon and 24-phase boards, and with AMD boards now getting 12 and 16 phases, and AMD releasing the 7371 16-core for high frequency, we might see the VRM done better and a new line added. On price, if they do the VRM better and allow PBO, I'll be doing a server workstation next round.

Sent from my SM-G900P using Tapatalk

 

So, here are some builds that are pretty good, moving up in price (prices as of time of posting):
8400 build ($835)
https://pcpartpicker.com/list/tDL7gw
1700X build ($1000):
https://pcpartpicker.com/list/T9dMKB
2700X build ($1088):
https://pcpartpicker.com/list/g2wDXP
8700K build ($1154):
https://pcpartpicker.com/list/76fVRJ

Except for MB, all use the same ram, the same mid-tier graphics card, etc. The Radeon 580 can be switched for a 1060 6GB, which is roughly market equivalents. The motherboards selected came from the recommendations on power delivery for overclocking made by buildzoid on his videos for GN covering the AMD Ryzen motherboards (B450 and X470) and Z390 motherboards. Earlier today, the 1700X was selling at $150, $25 cheaper. The 2TB HDD can be removed if not needed. But generally solid builds.

 

I'm pretty surprised by how quiet the x299 "refresh" has been. I have a 9800x that will be delivered tomorrow, I can't even find a review on it at all. I'd love to see some results before I open the box, lol.

 

Not alot out there...

Translate these:

Mother "X299 AORUS MASTER" compatible with Core i9-9980XE is released, made by GIGABYTE
AKIBA PC Hotline! Editorial department December 7, 2018 11: 00
https://akiba-pc.watch.impress.co.jp/docs/news/news/1157115.html

"Core i9-9980XE and other LGA 2066 motherboards compatible with the ninth-generation Core X processor appeared, and "X299 AORUS MASTER (rev. 1.0)" was released from GIGABYTE.

　The store front price is 47,304 yen including tax. The selling shop is Tsukumo eX.personal computer hall ,Tsukumo personal computer head office, personal computer shop arc, personal computer shop Akihabara BUYMORE shop"

With Intel Core i9-9980XE, Gigabyte X299 Aorus Master measured
By Chris.L on 2018-12-02
https://benchlife.info/gigabyte-x299-aorus-master-with-intel-core-i9-9980xe-12022018/

"...We chose to test with the Intel Core i9-9980XE with the X299 Aorus Master.

Hardware specifications, Gigabyte X299 Aorus Master is not a big problem, there are no shortage of specifications, while retaining VROC, Thunderbolt daughter card header, and upgrade Sabre DAC and Wi-Fi antenna, as well as the most important CPU power supply.

The design is worthy of praise; if you really want to pick this motherboard, the BIOS experience may be part of Gigabyte's need to work harder.

At this stage, Intel's Skylake-X refresh processor has not yet been seen in Taiwan, so users who want to change platforms may have to wait.

However, if the price of Skylake-X's 7000 series processor is adjusted, it is a good choice to choose the new X299 chip motherboard (Gigabyte X299 Aorus Master), just don't choose Kaby Lake-X processor."

 

It's mainly because this gen is same as last, just solder and tweaks to voltage. Basically the same as 7th gen counterpart, which is a bit ehh to cover. Not saying that makes it bad, just doesn't drive consumers to sites.

Sent from my SM-G900P using Tapatalk

 

I was hoping the 9800x would be more similar to the 9900k in all honesty. There was a lot of talk about 14++ but at least with the 9980xe it seems like it's just a soldered 7980xe. I may just make do with the 7740x that I have until Zen2 comes out so I can make a decision. I would like to have quad channel memory and the extra lanes, but I could get a 2700x and mb and save a couple hundred bucks from just not buying the 9800x, lol.

 

To be honest, it all depends on your needs as to what I recommend. First, I recommend waiting either way. Skylake-X was made on 14nm+, and this was a refresh for the 9th series. Intel is supposedly announcing cascade-X this month, likely at the Dec. 11 architecture day. AMD will be discussing Zen 2 at the CES as the keynote. Zen 2 should shake things up a fair amount.

But this is where your needs come in. If you need more memory bandwidth or PCIe, I'd say wait until next summer for TR3, if possible. If you need something before then, wait for cascade-x or for Ryzen mainstream, as the cost/benefit for respins isn't as appealing. If you need an immediate upgrade, buy a used 7 series 7900X, which then increases memory channels and PCIe lanes. It really comes down to what your needs are and if you can wait for an upgrade. I think CES may be the real goal, as speed and I/O of Zen may be closer to finalized. If nothing else, so long as everything goes fine for AMD regarding the keynote, the used Intel market may make the 10 core more reasonable (as well as hoping availability of the 9900K increases, thereby putting price pressures on Intel's 7900X, both in the new and used markets).

But that is my opinion. Hope it helps.

Sent from my SM-G900P using Tapatalk

 

I appreciate the opinion. I think we are both of a similar mindset. I actually sold my 7920x before the release of the "9th" gen because I thought the value would tank. I think I'm probably just going to wait a month and see what happens. Good point on hopeful intel price drops in the used market with the release of Zen2...

 

thats another issue here is the mindset of a lot of intel's loyal customer. hoping for zen2 to be decent enough so intel's chip would drop in price, then go for intel. honestly thats a normal mindset, sad to say this time around with zen2, AMD will have both value and performance crown if rumour turns out to be true. @ajc9988 12 or 16 cores cpu at 4.6ghz, assuming IPC is evenly matched with decent software support, i'd take that hand down over a 8 core intel cpu that can do 5ghz.

also that 9800x.. kinda a poor time to buy it, though considering you already invested the x299 board it'd make sense. im in the same shoe with z370 6 cores laptop.

 

I have no brand loyalty, I think you're missing the point of my post. I prefer performance and will pay a premium for it, but only to a point. If the Zen2 rumors are true then I will run AMD. If AMD is still behind in performance and it drops the price of Intel chips, I'll keep my x299 setup as I am happy with the board and setup in general. The biggest question is will the Zen rumors actually be true...

 

The Less you Spend, the More You Save - Not-Jen-Hsun-Huang

Not Buying Intel Save's Money - AMD

 

OMG, that is sooo Intel Brainwashed Zoomie Thinking.

AMD comes out with better CPU's than Intel, so hey, let's buy Intel CPU's at distressed prices instead of funding the company - AMD - that pulled us out of the 10 year 4c/8t nose dive Intel was taking us on.

" Duhhhh, let's give our monies ta Intel, cause we can't figure out WTF is going on?!!" - Clueless Intel Fan's everywhere

Sigh.

Help yourselves out of the wet paper sack you have over your heads. Stop buying Intel, stop giving them the power through capital, give money to AMD to effect change, to fund the change AMD has started so they can continue to do so.

That's why you stop giving money to Intel, because Intel doesn't deserve it, AMD does.

Spoiler: 10cm noose visualization...

 

He already owns the platform. If going for TR, if he cannot wait until next summer, just getting something on that platform is fine. I would hope the Ryzen chips release in March, but with the X570 boards coming at computex, there is a chance of waiting until May, which also may vary by SKU. In that regard, buying on the depressed pricing, if he cannot wait for release (I do not expect a full release at CES, rather an announcement). The difference is a matter of months. Who am I to say not to buy if the need is immediate? I expressed generally waiting will be good.

Sent from my SM-G900P using Tapatalk

 

Yeah, still though, you said it. Buy Intel CPU on good news from AMD CPU release...typical over excited Zoomie response...

You've seen doggo zoomies, same thing, mind is going fast - but with no place to go.

Besides, better to waste Intel setup components than throw good money after bad spending $2000 on an outdated CPU.

 

Lol I think ajc9988's comment is being a tiny (ok majorly) blown out of proportion.

 

Yeah, why think and shake things up?, that's never helped move civilization forward before... right?

Think and Save, invest in the future, not in the mistakes of the past.

So many little things are small in their apparent view and effect, when you can't see through the veil of opaqueness hiding their effects, repeated millions of times daily - normal isn't normal, especially if you adjust your point of view to take in a larger view.

The smallest decisions have fated man to failure throughout millenia.

Wash your hands? A small act that wasn't seen until it was obvious.

Stop giving Intel money.

Same thing.

 

Hah I never trust someone too far on either side of the fence. But great speech.

 

If it didn't reach through into your thought processes, allow you to alter your behavior for the betterment of yourself and others, maybe it will for someone else.

 

True, lots of sheeple out there in both camps.

 

You can't hope to change their minds, merely aim to wake them up and look around at the world anew, perhaps for the first time.

And, I don't dis anyone's reactionary thought's - we are all human - doing our best running our software through an emotional soup of conflicting sensory inputs.

I'm only pointing out the crucial turning point of thought loops that we can all enter into, and not have a break() out - throwing a not-while error interrupt is the best I can do at times.

Anyway, Starve the Cold Dead Heart of Intel, and instead Feed the Fever of Innovation at AMD...

 

Intel Core i9 9980XE Review - LEO DOESN'T like this TURKEY!
KitGuruTech
Published on Dec 8, 2018
Today Leo takes a look at the i9 9980XE - and comes away less than impressed with how it compares against AMD Threadripper solutions. It certainly will be the most expensive Turkey you buy this year at Christmas, thats for sure. If you have £2,250 spare of course. Ouch. Read full review over at www.kitguru.net HERE: Coming soon!

 

Guess we will see if I can resist temptation:

 

So this is a bit more than just a rebranded / soldered 7820x? The 7820x had a 140w TDP with an 11MB cache vs the 9800x @ 165w TDP with a 16.5MB cache - with a slight increase in base and boost, with the same 8c/16t core/thread count, and the same $589-$599 MSRP.

https://ark.intel.com/products/1237...-X-series-Processor-11M-Cache-up-to-4-30-GHz-

https://ark.intel.com/products/1891...-series-Processor-16-5M-Cache-up-to-4-50-GHz-

And, the 9800x is the same 8c/16t count as the 9900k with $488-$499 MSRP, or $100 cheaper for the 9900k without the "fake" shortage.

https://ark.intel.com/products/1866...K-Processor-16M-Cache-up-to-5-00-GHz-?q=9900k

I wonder how all three compare in unlocked performance / power?

Wait!, no!!, I do not wonder!!...do not enter...


Maybe your Marvel friends behind that box of doom can help you?

 

If the 9800x was as good as the 9900k is when it comes to single threaded and gaming then I would probably just keep the 9800x. The problem is no one has any hands on information on the CPU. Intel is also using the term "smart cache" on the 9800x, no idea if that is really any different and if it would help for the slower mesh interface vs ring bus.

 

do u still got your 7920x rig? if so help me test out some stuff, like TBM3.0. iirc TBM 3.0 is an intel pre-set value for it's cpu similar to TBM 2.0 and if thats true then it sucks..

but if it isn't, can you do overclock say 4.5ghz on all cores, and set in XTU/TS for 50x for 2 cores. so it'll look like 50 50 45 45 45 45 45 ... 45. then run cinebench for example, single threaded first time without affinity and 2nd time with affinity pin to those best core that does 50.

 

I no longer have my 7920x. I could hit 4.5ghz on all cores, but under prime95 it was more than my 280mm AIO could handle.

I never used TBM 3.0 or any others, but as far as I know it's just the same preset clockspeeds. I don't know if it actually determines which cores are better than the others. I do like that AMD points to the top cores so you have an idea what to set them to.

 

manual OC is always better because you have the choice. @ajc9988 does AMD's oc software allows individual core voltage adjustment? iirc on x299 they allow it so say 7920x all core OC 4.5ghz at 1.18v for example means each core at 1.18v but since some cores are better than others, people can lower individual core and test for stability, though its very time consuming, it'll give the best temp results. does AMD currently have that? i'd love to have that come with zen2

 

On the 1950x, no. Do not know if they added it on the Zen+ chips, but do know they are able to read a per core voltage.

http://www.asrockrack.com/general/productdetail.asp?Model=EPYCD8-2T

Motherboard worth checking out. The giga board for single socket kinda had the memory in the way if putting a graphics card in the first couple x16 slots.

Sent from my SM-G900P using Tapatalk

 

South Korean Ryzen 3600X and 3700X contest is pulled after AMD is notified of its existence notebookcheck.com | Now 11

Steve Walton of TechSpot and Hardware Unboxed reached out to AMD for comment and was told that the contest was not sanctioned by AMD and that it has now been pulled. He also made several remarks on the recent rumors from Reddit and AdoredTV, stating

The Ryzen rumours are fake, pretty certain at this point. I've got info off the record so I can't released [sic] it, but yeah people are going to be upset when they really shouldn't be... don't assume Ryzen will be anything like Rome.
​

If Walton's statements are to be believed then the actual CPUs won't quite match the high expectations set by recent leaks but will nonetheless be exciting. Some may have thought this contest was proof that Ryzen 3000 would be launching at CES per the recent rumors, but now that the contest has been pulled, it is questionable whether or not this sales agency was aware of either the Cinebench scores or the release date(s) for these two CPUs.

 

i'd personally say adoretv's speculation is too optimistic with the clocks. but i believe with how zen2 is designed, with 7nm as it's target for power consumption, a 12 or 16 core is very possible even if its at low stock clock say 3.5ghz.

since it is easy to discredit others, techspot and hardware unbox will need to come up with some of their own analysis why adoretv is wrong and why rumors are fake, rather being told "not real"

 

Try looking at Der8auer, who said he expects 5-7, up to 10% over current Zen+ on speeds. That is like 4.5-4.7, depending on chip, roughly in line with my own prediction. Also, boost, as of Zen+, was single, not all core. I feel like a broken record at times.

Edit: so to add on this, didn't you agree with Talon in his basically saying Hardware Unboxed was unreliable in the 9900K thread, now referencing what was said?

Sent from my SM-G900P using Tapatalk

 

What are you talking about. All core on Zen 1 was 3.7 or 3.8 GHz. Zen+ 2950X was around 4-4.2GHz, depending on certain things. 5% on top of that is 4.2-4.4, while 10% is 4.4-4.6. Then, take into account a different platform, so more thermally dense than TR. So saying 3.5 is low, unless talking base, and even then seems a bit low.

Edit: exact specs of the 2950X are 3.5GHz base, 4.4 Max boost (single core), no all core boost because of precision boost/overdrive. 4.2 was what was seen on all core, whereas 4.0 was decently common on Zen 1. 0.05x4 = 200mhz, 7%= 300, 10%= 400.

Sent from my SM-G900P using Tapatalk

 

So, wanted to add on Intel saying that 7nm is on track. Here is an article on that timeline (spoiler: late 2021 expected for Intel EUV):

Intel Ceding Leadership in EUV
By Alan Patterson
https://www.eetimes.com/document.asp?doc_id=1333657
http://infozonic.com/2018/12/10/intel-euv-enabled-7nm-process-tech-is-on-track/

Now, today is Intel's architecture day, an invite only event. I have seen nothing to suggest a live stream (which Intel's and Nvidia's live streams of late have been fairly horrible, but AMD did not live stream the Nov. 6th event, instead posting it in its entirety on youtube after the presentation was over (think it took 1 day before it was uploaded)).
https://hothardware.com/news/intel-sends-exclusive-invites-for-next-weeks-architecture-day
https://www.tweaktown.com/news/64022/intel-hosting-architecture-day-2018-december-11/index.html
https://www.techradar.com/news/intel-architecture-day-2018-promises-talks-about-next-gen-cpus
https://wccftech.com/intel-architecture-day-2018-announcement/

But, lest we forget with the alleged leaks putting two of Intel's Cascade-AP 48 core chips versus two of Intel's Rome 64-core chips and winding up within a couple hundred points of each other on Cinebench, that it is rumored Intel's 48-core chips REQUIRE liquid cooling, something AMD does not and something that cannot be deployed widely in data centers due to risks of leaks.

https://hothardware.com/news/amd-64-core-rome-epyc-48-core-intel-cascade-lake-ap-xeon-cinebench-leak



Note: The Intel Cascade-AP was running at 2.5GHz whereas the AMD chips were at 1.8GHz. Meanwhile, AMD showed off around 2.35GHz at the press day Nov. 6th, and HPE showed in a slideshow 2.35GHz, meaning that this may not even adequately represent AMD's offerings, as 2.35GHz is a 30% increase over the 1.8GHz here.

https://www.semiaccurate.com/2018/11/14/coolit-water-cools-cascade-ap-cpus/
https://www.anandtech.com/show/13598/amd-64-core-rome-deployment-hlrs-hawk-at-235-ghz
https://www.tweaktown.com/news/63827/amd-epyc-rome-64c-128t-processor-teased-2-35ghz/index.html
https://wccftech.com/amd-epyc-rome-64-core-7nm-flagship-cpu-clocks-2-35-ghz/

So, the server battle is heating up, potentially with Intel being too hot to handle (without water cooling) while AMD is sitting in the breeze. Should be some good articles coming out today, so...

 

i think you misunderstood what i said. im talking about power consumption in my paragraph.

also, serve the home showed CB getting broken by 4 older xeon chip and score is unstable, so that leak of CB test with AMD can't really be trusted at this point as it could be fake. either that or they could run say 10-20 times and pick the highest score out of it so it could still be possible, but you get my point.

 

Yeah, reread. Just find it low balling, so to speak. At 135W, that is a reduction of 45W, which is a 25% power reduction. Since it's 50% reduction or 25% performance, I could see 12.5% or slightly more going toward performance with the 25% power reduction (not perfectly linear, so not sure the performance uplift is exactly half). As such, and since the 1950X could get 3.7 all cores in power envelope, even if the performance uplift is all from IPC, all core clocks would likely be around zen 1 or zen+ all core boost.

As to what STH showed, it only broke AFTER an Intel firmware update to the microcode and has not been shown to be broken in any other situation. This is why I'm of the belief Intel's microcode was broken instead of the benchmark.

Sent from my SM-G900P using Tapatalk

 

So all Intel has is an active interposer with foveros, which is something AMD has planned since 2014 but the cost analysis white paper of 2017 said cannot do it yet. See adoredtv videos on interposers, chiplets and butter-donuts and follow up video.

Sent from my SM-G900P using Tapatalk

 

Looks like they have 10nm coming next year to consumer desktops and 7nm has been reported by them as progressing better than their expectations. 2019 is going to be exciting.

Also the 3D chip according to one article I read has been in development longer than Intel has been working on 10nm. It seems Intel has figured out how to get it done, probably because they actually make chips themselves in house.