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

That top one looks like even form factor won't be a problem. Sweet.

 

Intel vs AMD 2018 - AdoredTV 2018 Musings

 

Supercomputing 2017: Cray ARM Supercomputer with Cavium ThunderX2
http://forum.notebookreview.com/thr...m-supercomputer-with-cavium-thunderx2.810913/

 

Awards: Best CPUs of 2017 (Gaming, Production, & Biggest Upset)

 

intel 8 cores im waiting, 14nm++ or 10nm+, either one will need a new laptop redesign from clevo/msi otherwise 1 FAN cooling? better be a fat heatsink as large as gt83vr GPU

 

https://www.myce.com/wp-content/images_posts/2016/06/intel-octane-ssd-roadmap.jpg

this road map was so long ago i hope its still accurate. intel mentioned they'll start selling some ULV CNL mobile devices end of 2017 and more products into Q1 2018. we can take that Q1 and turn it into possibly Q2 or near end of Q2.

if roadmap hasnt changed for CNL platform then we can expect to finally see a decent size m.2 optane SSD pcie 3 x4 lanes for laptop, pair that up with 8 cores 5ghz from intel w00t

 

They have to fix this first... “Power consumption on the Intel Optane SSD 900P is rather high, which is why you don’t see a M.2 drive as this isn’t going to be something you want to stick into a laptop. This drive idles at 5 Watts and has a burst power of up to 14 Watts. The average active read power is 8W and the average active write power is 13W.”

 

unsure where most of that power came from, its likely the optane flash/memory rather than its controller. because by CNL platform it'll have 2nd gen or possibly 3rd gen of x-point memory it'll likely have better efficiency thus reducing power cosumption. also 900p is 280gb and 480gb and more memory = more consumption, we just need a 100-120gb m.2 is enough and im sure they can get it to work.

we already got a CPU that use easily 15x more power in a laptop, a 100gb m.2 optane SSD probably wont use more than 8w, which imo 8w still comparable to samsung 950 pro back then which i think average active peak was around close to 7w but of course that is 512gb.

 

But you forget that same Optane drives shall be used in thinner and even more flimsy. They can't push out those drives only for normal sized laptops Heat and power usage will be a big problem. And a lot of the laptops/tablets havent't much storage slots either. Put in a single 100/120GB Optane drive and the laptop is almost fully up

 

2 in raid

optane cache might be used in thinner laptop but if its 100gb+ then its optane devices will be used like an SSD, it'll have high power consumption. and honestly speaking you know as much as i do that we dont care how they are used, as long as its not a BGA piece glued onto mobo, we can easily place them in our high end laptop. pricing obviously will be high, capacity low, thats just what the new optane technology really is, but performance will be tremendous.

znand wont be ready anytime soon, certainly not the upcoming 980 pro in a few months time. and even if znand has high 4k read and low 4k latency, it is still nand and it's performance gets destroyed during a mixed type of workload and during steady test state. optane also gets faster as module is smaller, in terms of 4k latency but suffers sequential performance due to less channel.

http://tssdr1.thessdreview1.netdna-...s/2017/11/Intel-900P-PCMark-8-Extended-AB.png

 

@ajc9988 @tilleroftheearth @Papusan any many other who i missed

https://twitter.com/FanlessTech/sta...io/iframe/twitter.min.html#934109005693284357

so its confirmed eurocom F7 will likely be getting 8c icelake 10nm+, not saying there might not be an 8 core CFL on 14nm++ but if they gonna have 8c 10nm+ end of 2018 they wont bother spending more money on this. CFL 8 cores 14nm++ with z390 in a few months time seems unlikely when they can do 8 cores 10nm+ on z390 end of 2018.

@Papusan this 8 core 10nm+ might reduce temp by quite a bit going from 14nm++ to 10nm+. though 14nm++ is a better silicon quality but the shrink itself will reduce consumption under most usage scenarios.

 

It can't come fast enough, but at the same time, there is still nothing better now than what I'm running for my needs.

 

AMD: Ryzen 3 1200 delivered with 8 active cores

"In Russia, apparently several cases have emerged in which an AMD Ryzen 3 1200 was delivered with 8 active cores, although this should actually have only 4."

"Just a few weeks ago there were reports that apparently Ryzen 5 processors with 8 active cores + SMT were delivered, although actually only 6 cores should be active. Now there are apparently similar incidents on some Ryzen 3 processors sold in Russia."

 

Now that is a lottery I like!

 

Intel Core i9-7900X 3.3 GHz Review-Techpowerup.com

Value and Conclusion
"Intel's "Skylake" micro-architecture continues to enjoy higher IPC than AMD's newer "Zen," and it reflects in higher performance than the AMD Ryzen Threadripper 1950X in every less-parallized benchmark, including games. It's only at GPU-limited resolutions, such as 4K Ultra HD, that Threadripper catches up because the GPU is the limiting factor. The 7900X's single-threaded performance, however, isn't higher than that of the mainstream-desktop Core i7-6700K despite the higher L2 cache (refer to "CineBench R15 ST"). The idea behind more L2 cache appears to be more one of reducing latency in inter-core communication than an attempt to improve IPC."

"The Core i9-7900X trades blows with both the $799 Threadripper 1920X and the par-priced Threadripper 1950X at multi-threaded tasks. In most synthetic multi-threaded benchmarks, the 1950X is more than comfortably ahead of the i9-7900X due to its six extra cores. On multi-threaded media encoding tests, such as x264 and x265 video-encoding (which is a real-world test where we're feeding a test video to the encoder), the i9-7900X is ahead of the 1950X."

"If you work with highly threaded apps all day or have full control over the code your software runs, which means it can be hand-optimized, Threadripper could be a good alternative due to offering more cores (= more performance) for the money, but such scenarios are fairly limited. Also, if you do 4K gaming, which will be GPU limited anyway, then Threadripper could be reasonable if you want to keep up with Game Mode switching."

 

That is a good reason for AMD to keep the 1950x at $800 or below.

 

Gaming And Streaming: Which CPU Is Best For Both? - Tomshardware.com

"Many streamers place video quality over maximizing the frame rate of whatever game they're playing, so your own priorities will largely dictate how you tune your system. In fact, turning on v-sync may be a good way to balance streaming and gaming performance.If you seek the highest in-game performance while you stream, Intel's Coffee Lake-based Core i7-8700K is a good fit. The Ryzen 7 1800X is also competitive and tends to offer better streaming performance. Using our settings, the 1800X also had more CPU headroom leftover for more taxing encode settings, if desired. Granted, some of that extra horsepower is due to the 1800X's lower gaming performance, which means there are fewer frames to encode."

 

iMac Pro vs. Alienware Threadripper!


Insane $7,200 price on that iMac!!

And, while rendering it only just matches a Ryzen 1700x build in comparison, and the ThreadRipper AW build is much faster.

I'm sure an NBR member could build both the Ryzen and ThreadRipper builds for far cheaper and OC / tune them to way out perform the iMac and these commercial AMD builds!!

 

Benchmark wise my $4,000 TR build demolishes the AW machine from reviews I have seen. Now that the TR is reduced in price it could be done even cheaper.

 

Alienware is all about design. Not raw performance.

 

As a Software Dev I switched from my 8700k to a Ryzen 7 and could not be happier.
https://www.reddit.com/r/Amd/comments/7oqp93/as_a_software_dev_i_switched_from_my_8700k_to_a/

"After a long love/hate relationship with Intel over the last few years and being frustrated with with what’s happened this year and last, my software dev and personal home machine have switched over to AMD.

I originally just about 2 weeks ago built an 8700k version of the exact PC I’ve listed below, but after patch and the performance hits I took in Virtualization, compile time, NVMe speed.. I’m done with Intel.

It was enough of a hit for me to call up my local store and get an exception/return period bump so I could actually swap to AMD."

Amdestroyer94 AMD 91 points 14 hours ago

"How is the performance when compared with your previous system. Also what was the degree of performance degradation with the patch that made you switch"

HugeIRL Ryzen 7 1700 | 1080Ti | Corsair Fanboy[ S] 109 points 14 hours ago

"The Ryzen 7 1700 is a little slower performance wise but mainly due to clock speed. It still does everything I need and the extra cores/threads outweigh the clock loss for me (I plan on hitting 4GHz soon, right now I’m cozy at 3.8/1.350v).

For me it felt drastic. I saw compile times drop significantly especially with Java, my NVMe drive was a touch slower too leading to my large transfers suffering.I felt I easily hit the upper numbers of degradation (30ish%) especially with the server tasks/virtualization I was doing.

I game in 4K as well so it’s not .. much slower? It’s within margin of error for that since I’m not CPU bound at all. I’m coming from a 1070 build from last year(to the 8700k and then the now new Ryzen 7) so it’s a huge step up for me."

 

Did you see Papermaster's slide on tech day showing that the 12nm they are comparing to 14nm+ (Kaby) on the process side?

 

I've heard the same thing from several people, getting the impression that virtualization is getting the absolute worst of it. Though oddly my earlier prediction of it also having a major effect on large databases has been pretty quiet on that front.

I mean, I see performance degradation, but nothing like what's happening with virtual boxes.

 

Not specifically, do you have a quote, link, slide?

 

You quoted a quote from an article I quoted, that wasn't my personal experience

I have seen early reports from the postgres guys about large hits in performance:

heads up: Fix for intel hardware bug will lead to performance regressions
7%-23% transation performance penalties for Postgres with PTI patch.
https://www.postgresql.org/message-id/[email protected]

And there have been many reports from AWS customers with high levels of performance degradation, but they don't mention specifically what their servers are doing, I would imagine VM's and DB's among other services based instances.

CPU Vulnerabilities, Meltdown and Spectre, Kernel Page Table Isolation Patches, and more
http://forum.notebookreview.com/thr...page-table-isolation-patches-and-more.812424/

 

It was a slide from the AMD Tech Day that it seems no one wants to share. Go to 16:10.

 

Well, yes, I've seen that slide, I call it "narrowing the IPC gap", is that your point, the shrink and the improvements from AMD are narrowing the IPC at the same speed?

 

I don't consider it IPC, per se, as it is narrowing the process efficiencies. Other aspects, such as uARCH effect the IPC, as do the changes to latency of the interconnect. It also doesn't speak to the speed achieved on the process node, which people often lump altogether when they talk about IPC. But, roughly, yes! I am excited they are jumping generations at a time for Intel's incrementalism to now be closing the performance gap so quickly that the better, or more scalable, uARCH will win with the processes being the same, with AMD already showing SMT to be better than HT on many aspects.

 

Sigh, I finally read the axis label, it's Performance Per Watt, not IPC. That also makes sense.

 

AMD Announces Enmotus FuzeDrive technology to Speed Up Ryzen-based Systems-Techpowerup.com
"AMD today in a blog post announced the fruits of its partnership with Enmotus, a mainly enterprise-focused company that has made its name in creating performance-optimizing software solutions."

"L; DR: Essentially, the performance increase doesn't come from activating the FuzeDrive software on the Ryzen system; but from both activating it and adding a Samsung 950 Pro SSD to the system. How much of the performance improvement can be attributed to the added NVMe drive alone is a valid question, and we'd wager it's most of it. You can purchase a license to Enmotus' FuzeDrive for your AMD system for $19.99."

 

Windows 10 Meltdown-Spectre patch: New updates bring fix for unbootable AMD PCs
AMD PCs can now install Microsoft's Windows update with fixes for Meltdown and Spectre and the bug that caused boot problems.
By Liam Tung | January 18, 2018 -- 14:02 GMT (06:02 PST)
http://www.zdnet.com/article/window...new-updates-bring-fix-for-unbootable-amd-pcs/

More details in this thread:

CPU Vulnerabilities, Meltdown and Spectre, Kernel Page Table Isolation Patches, and more
http://forum.notebookreview.com/thr...patches-and-more.812424/page-63#post-10666767

 

AMD Ryzen 5 2500U Vega 8 vs Intel i7-8550U UHD 620 ...or why there is Kaby Lake G

 

https://www.reddit.com/r/intel/comments/6x2mj9/turbo_boost_30/

good thing i didnt buy a desktop. early tester wowzer, TB3 something i always wanted, more cores too. but the mesh clock is a joke, so many things needs to be fixed right now on HEDT before i consider it. crippled performance in many way.


btw this is 870TM with 8700k, its beating ryzen 1800x pretty nicely.

 

Good for a 6-core, but you have to hit 4.8 to fall short of a 4.0-4.2 overclock on an 1800X. So what are you talking about? Plus, your ram timings look like crap. Tighten those, then let's see the score, which should go up potentially to mid 1700s. But, to be honest, unimpressed. Now, the fact you do not need 33% more speed to hit that, rather closer to 20-25%, is nice and shows where the difference in structure performance is on IPC. But that really is tedious bro!

http://www.hwbot.org/submission/3693710_cox_cinebench___r15_ryzen_7_1800x_1947_cb/



Edit: Here is 4.0 with slow ram timings on 1800X That should be more comparable.

http://www.hwbot.org/submission/3720112_justifiedparanoia_cinebench___r15_ryzen_7_1800x_1644_cb/
It is only ranked 55th out of 60 on that chip on HWBot for that benchmark. Kind of says a lot!

 

try one more at 2400mhz i wanna see close comparison, or i could just go by the math difference but that woulud be in accurate.

also, im talking exactly that. 4.8 is a normal clock for 8700k, actually considering thats 100% of SL chip can do that. but ryzen chips at 4.0? some can only do 3.9 and the high voltage past 4.0 isnt really worth the extra power consumption.

basically i got a 6 core chip to match a 8, while blowing away at single threaded performance, pretty damn amazing id say.

 

I pulled both from HWBot. Look at the links. I have a 1950X. You are barking up the wrong tree.

And, that is an overclock. On an 1800X, almost all hit 4.0. Some to 4.2. If you said 1700 or 1700X, then 3.8-4.0 was much more common. Some issues were helped with firmware and some with people learning how to OC AMD, which has different techniques that are similar, but not identical, to Intel. In fact, the on air average on HWBot is 4044, while the water avg is 4077 or 78. That means what I posted was a good chip that smoked yours and an average chip with low ram speed (2660).

With the 8700K, 5.0 is the air average and 5.1 is the water average. But, that also means 25% comes from speed, with the rest on IPC and other structural differences. But in a raw sense, it all comes down to your purpose for the machine and what software is used. We've gone through this a million times already, yet you beat a dead drum.

And, considering you have a chip Intel unloaded instead of eating inventory that has a known built in exploit at the time of sell, I really have to say you got fleeced! Especially getting it AFTER it was known about. Sad, really. Now, with that said, that chip was originally planned to be released this quarter, meaning it was meant to go against the Ryzen 2 2800X. If you add the approximate 10-15% performance onto those scores I posted above, with a likely 4.4GHz OC, then we are talking about a push, really. So, we will see that in April. We then get Intel's 8 core mainstream really coffee this summer, which should improve more and take back that space, but we'll see cascade vs TR2 at that point, with all the arguing around that, when none of that matters as Intel falters with Ice the following release and Zen 2 7nm really lands. Need I go on?

 

woah your old bad habits showing again! calm down mang, deep breathe.!

 

What the hell are you talking about? I showed the extra OC headroom of the 8700K in a desktop build, that way to show the higher speed and scores. I point to an overclocking database of scores which is known to be reliable, while we all understand limitations on laptops. Then I put the releases in perspective of original timelines, adjusted timelines, and why comparing it to the new vs year old chip is what is proper due to Intel's impropriety. So, do you not like being called out for that? Or what?

 

TBH that 8700K should do much better than what is shown.

 

no no, what the hell are YOU talking about. running 4ghz in a laptop for ryzen 8 cores, it'll overheat like crazy. this at 4.9ghz 24/7 is awesome and thats below 80c on auto fan. i could crank up the fan by another 20% for 5ghz 24/7 and stay under 80c but i'd say the extra noise isn't worth it. oh right, btw this is including avx workload which people wanted -2 normally to maximize sse but naw thats too weak.

 

You are running near 80c 24/7? I give your cpu about 6 months before it craps on you. Ryzen 8 core at 4 ghz, even in a laptop, wont run near that hot. Ryzen 8 core at stock clock with no heatsink or fan idles at 65-70c. Ive seen it on my Ryzen.

 

Intel CPU's are a lot sturdier than that

I've had CPU's running 24/7 for long stretches in production for years, and temps aren't always that low... in large facilities you can get things cool enough, but not too cool, otherwise it costs too much in cooling.

Staying under 90c is "good enough" to get long life for most CPU's, it depends on the process used, AMD has lower than Intel temperature extreme limits, so maybe for AMD process it needs to run lower temps.

The high core count CPU's with high OC and 250w-850w CPU package power draw, just aren't going to run that cool without a lot of effort. Air-cooling isn't going to be enough, but many will be air-cooled only and are going to run hot.

Intel's have been burning hot for many years, living long lifetimes.

 

high temp means chip require higher voltage so yeah but 1 year seems under whelming. i have had many chips overclocked and hitting over 80c with higher voltage than i currently use and they have lasted years, i'd say no problem.

 

For you my friend
Impact of Temperature on Intel CPU Performance-Pudget systems
"For the average system, our rule of thumb at Puget Systems is that the CPU should run around 80-85 °C when put under full load for an extended period of time. We have found that this gives the CPU plenty of thermal headroom, does not greatly impact the CPU's lifespan, and keeps the system rock stable without overdoing it on cooling. Lower temperatures are, of course, better (within reason) but if you want a target to aim for, 80-85 °C is what we generally recommend." But when you cross the border... 85C, newer cpus can and will start experience instability if they run with a decent overclock.

Exactly.

 

Okay, don't gang up on me

Sent from my SM-G935T using Tapatalk

 

Wasn't the meaning. The Info was not only meant for you. To you all

 

I know I forgot Intel since the P4 days have been like blast furnaces.

Sent from my SM-G935T using Tapatalk

 

Ryzen 7 1800X vs. Core i7 8700K, Meltdown & Spectre Updates Benchmarked

 

90 still seems high, I get antsy about it if I go over 80 to be honest.

 

Under 90 means in the 80's under load, that's good enough to run for long periods of time, if you want it lower run 100% fans, but for me that's too noisy