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

http://www.tieba.com/p/5352798808
so.. clevo just using the same chassis the new CFL 6 core. which means only mobo/chipset updated on the mobo and very likely to design same size in order to reuse same chassis, heatsink, etc etc.
the current heatsink can't cool off 4 cores at 5.2ghz let alone 6 cores. gg clevo cheapin out.

well some good news for current owners they can simply replace mobo + cpu to upgrade to 6 cores its very likely possible.

 

Wow, time to order that RX 64 Liquid Cooled model...

 

@ajc9988 @hmscott @Papusan @TANWare @tilleroftheearth

heres some more from hardocp which is from baidu..



rofl 9980x? 9700k lololol. 10nm++ looking damn strong man, 8 cores in a laptop see that 9750k at turbo 5ghz?? possibly the devil canyon version of 9700k. waiting game begins.. i want the next gen 9700k 10nm+ then i can upgrade to 10nm++ with 9750k in a laptop.

8700k can later upgrade to 8750k which is 10nm but still only 6 cores.. YEAH INTEL MILK

 

Intel "Coffee Lake" lacks year-round availability close to AMD's next generation-Sweclockers.com
"The launch was originally scheduled to take place in early 2018, but should have been moved in a quarter due to AMD's launch of the Ryzen Processor family . Now, sources of SweClockers speak that the modified plans have an impact on availability and that Coffee Lake appears to be a shortage of the year."

"It feels like Intel is shooting itself in the foot with the launch of Coffee Lake. Coffee Lake will kill the sale of Kaby Lake, nobody will want to buy Kaby Lake. It's like Intel's only "releasing" Coffee Lake to try to curb the sale of Ryzen."


AMD Ryzen 5 1600/X CPUs With 8 Working Cores Spotted In The Wild-Wccftech.com
"Several recent Ryzen 5 1600X CPUs owners built new systems and upon booting to windows for the first time were shocked to discover that they had 8 cores and 16 threads. The CPUs had the same 3.6GHz base clock and 4.0GHz Turbo of the 1600X, but had 8 cores and 16 threads instead of 6 and 12 respectively. Making them effectively price-cut 1800X CPUs."

"Why exactly this is happening is unclear. Each CPU is tested right before it’s packaged to be shipped, so it’s implausible for the manufacturing facility not to have known that those 1600s and 1600Xs they were shipping weren’t 1800Xs. It’s possible that this was a decision made simply as the result of higher demand on Ryzen 5 1600X and 1600 CPUs leading AMD to use some of its less purchased 8-core dies as Ryzen 5 1600Xs and 1600s to meet the demand. One thing is for certain though, this gives a whole new meaning to the phrase silicon lottery."

Increased competition leads to unexpected solutions

 

well the leak from baidu could likely be fake take it with lots of salt at 100kg lol.

spelling error, 10nm+ and 10nm++ too far out, memory speed validation only comes when cpu is out/made, missing cascade lake which is SKL-X refresh in the chart.

 

Actually, @TANWare this seems to be two months, with the second month showing increases in the mainstream X series CPUs. There seems to be a slight decrease in number of CPUs sold on HEDT, but the revenue generated by the HEDT stack seems to have stayed the same. Now, the 7700K sales being down makes sense with the upcoming coffeelake, but other than that, it shows a trend to a degree. That is almost an entire quarter of growth in sales.

Meanwhile, due to cost of Intel's halo 7980XE, I think you overestimate it's desirability, although the 12-16 cores may have been waited for, hence the reference to the 1950X selling double the units of the 7900X. But considering the aggressive pricing on the 1600X for the upcoming coffeelake, I'd be surprised if it doesn't continue and that AMD doesn't adjust as needed.

NO! THIS WILL LIKELY NOT HAPPEN.Look at the first generation of the Phoenix versus the second and third, then the 270 chipset. The early chassis was changed enough to prevent compatibility and prevent GPU upgrades which pissed people off. Why would we expect the layout to be EXACTLY the same when Clevo has a history of making minor tweaks to make MBs incompatible with the next chassis, even if they look the same. Trust me on this. I went through it on the HM and the ZM, where, without a dremel and other mods, you could not do an EM or DM board in those chassis. So wait for more guidance before getting on the hype train.

This isn't too far out, as they can set performance targets on the upcoming chips and Ice Lake is already taped in (see prior posts describing the difference between taped in and taped out). They have them in development, but this is likely an aspirational chart. Meanwhile, 5GHz boost isn't as impressive as 7nm from GloFo is also targeting this. So let's back off the Intel hype train for a second. All this chart shows is cache, which can change late in the game as seen by Skylake-X, core count, which is speculative and in response to AMD, and speeds that are the goals and also do not separate boost on single core from all core boost, a main complaint about Intel as it is slightly deceptive. I personally wouldn't wait for **** from Intel. They have delayed EVERY 10nm PRODUCT TO DATE. They are practically killing cannonlake or planning on cannibalizing sales with Ice Lake on the turn around, etc. 10nm++ isn't until 2019-2020, going up against Zen 3, which means it doesn't look that strong, TBH. I'm unimpressed overall with their offerings because AMD got back in the game and Intel is stuck on a looser node going 3rd gen 10nm against 2nd gen 7nm. Logic density will be equal, but the 7nm likely will have better energy consumption AND reach higher speeds due to difference in the node and licensing from IBM. So who cares about that layout?

 

Also, missing DDR5 on Tiger for HEDT, which last time Intel made dual support on the MB and chips for DDR3/4, which means less likely. Also, Ice Lake isn't until late 2018 to early 2019, and Tiger is 9+ months later, meaning the timeline is complete BS.

 

I wonder if they are massaging the results by testing with games that have a lot of onscreen AI behaviors and physics objects. You could make those look like they've improved a lot independent of the GPU.

 

Unimpressed or not. The chips WHO give you the best performance and OC possibilities beat it. Aka all I care about. I expect Intel still will offer best performance for all(Amd best on prices). Anything else would be strange.

 

Lmao! Then you may need to get ready for the weird!!! Do research into the 7nm+ process versus 10nm++ and you might see why I'm unimpressed and doubt Intel's performance dominance in 2-3 years.

The price/performance is to get back in the market, moving forward, Intel is broke. They take too long per node. TSMC just announced when they expect to be on 3nm, with 5nm starting in 2019 and ramping in 2020. GloFo is looking at late 2020-21. Coincidentally, Intel won't be shrinking to 7nm until then and are abandoning full node shrinks with 10nm. Just saying, it looks like Intel hit a wall and everyone else is set to move ahead...

 

Intel is milking all the way and we know it. But they aren't that stupid and let AMD win the performance race(If AMD can offer best performance + best prices... = Intel will be as AMD was the last decade). Even with the information you provide, nobody knows what happens behind closed doors.

 

I think they are stupid. They've grown lazy and fat. The only benefit they have is regulatory bodies in the U.S. aren't doing anything for enforcement at the moment. But, with the past decade and recent studies showing a loss in brand loyalty, focusing on other factors than just brand, Intel will have a WAY harder time than a decade ago at getting over on consumers. Where AMD wins or comes close is covered better than a decade ago, with journalists being called out a lot more for BS spin!

 

Seems right to me. They'll wait until they have to change something and then they will. Sort of like how NVidia is not bothering with Volta to consumer cards yet because Vega is not much threat to Pascal on the top end.

 

Last time I checked,1070 and 1080 ARE the high-ed.
Also, pro software disagrees with you on the Vega not being a threat to Pascal on the very end of the 'top spectrum'.
Plus, recent Forza 7 DX12 benchmarks saw Vega go past Pascal 1080ti (by 23%) at 1080p and 2k... in 4k though, the1080ti won yes, but by less than 10%.

Forza 7 results:
https://www.tweaktown.com/news/59337/rx-vega-64-beats-gtx-1080-ti-dx12-forza-7/index.html

Compute benchmarks:
https://techgage.com/article/a-look-at-amds-radeon-rx-vega-64-workstation-compute-performance/

Given all the features and compute hardware inside Vega, games with proper optimizations to take advantage of those features should allow it to top 1080ti in games.
Plus, AMD said something about Infinity Fabric in Vega not being optimized for games... so we could see another boost from that (maybe around 10% mark - but this number is admittedly hypothetical as I'm extrapolating based on what high RAM speeds did for Ryzen's infinity fabric - really depends on how AMD optimizes "IF" here).

High power draw can be fixed with a simple undervolt (since we know AMD's voltages are not optimized, unlike Nividia's).

Vega is a new architecture that devs didn't really optimize for (sans Forza?) and is compared to Nvidia in mostly NV optimized titles... and those games that are optimized for AMD to date have not yet released patches for Vega's architecture.
Pascal had what... a year to be optimized for?
Polaris and Vega aren't the same uarc, and Vega has double precision math and other things in it that could make a difference in performance.

A decent comparison would mean to wait until the drivers are more mature and devs decide to release patches or new games that are properly optimized for Vega so we can gauge it's performance... but they should also optimize for Nvidia's Pascal too so we can see the differences in features and how much of a performance difference is there.

 

afaik DM changed the bottom plate and vaper heatsink into DM2/3 which you only need to replace bottom cover and thats it.

and this new CFL in 870km looks exactly the same i doubt they will improve heatsink and reason is obvious, money. though for sure i want them to up the CPU heatsink but its just day dream at this point. even eurocom kinda confirmed the outside look is basically just 870km.

 

It took more than than changing the bottom cover. The bottom chassis had to be changed. That is why Mr. Fox sold his DM to get a DM3.

 

Changing the bottom cover was needed for the cooling. The bottom cover for the change in ports.

 

And what is the likelihood they left all screws in the same spot so it is simply changing out the MB here? Especially with all the little changes Clevo usually makes?

 

Read the P870DM-G upgrade Guide

 

Which is replacing the majority of the chassis. You practically piece out your upgrade, which quickly negates the benefit over just selling the old one and buying the new one. Granted, you don't have to buy the graphics cards (unless 980N original design) or the display parts, but you recoup that in the sell.

 

The bottom line(end results), is what will be cheapest for you. But almost no point if the gain is too small.

 

There's theoretical parity but right now but I'm not seeing it make a difference any time soon. Waiting a year will do nothing here because if there is any actual closing of the gap NVidia is probably just sitting on Volta to take the wind back out of AMD's sales, pun intended. That it can currently outperform in a game and has future potential is overshadowed by it not being competitive right now, and the likelihood that by the time it is, what it's competitive against will be old news.

 

I disagree. Even with higher power consumption, because of low mGPU support, it may make sense to buy the Vega 56 when the price is back down to $400. The reason is, if you are shooting for 1080 performance on a budget, you get it, it matches readily the 1070 Ti, and you may get 1080 Ti levels of performance in some future games. We know Volta is coming, likely in March next year. If you can wait 6 months, then wait. If not, you have a great value option with future performance coming. The 64 is a harder sell, but if you are buying at that price point, the point remains, if the price is $500 or so, you have future performance uplift to around Ti levels, current performance of a 1080, and a nice price.

Now, the sad news is the lack of a win at 4K, although at 8% lower than a Ti, it still beats the 1080 on value proposition. So it is a matter of if you can wait or not.

I'd recommend, so long as the power supply won't blow the budget, a Vega over the 1070/Ti and 1080 once prices are around original MSRP. If you are buying a 1080 Ti, buy it. But, as always, if you can wait, do so as better is always around the corner. But if having to buy now, Vega does seem better situated for the future.

 

A fair point, and I was mostly talking about 64. The interest I see for both cards, if you filter out "I would literally install a rock in my computer if it didn't say NVidia on it" is mainly focused with cryptocurrency and other non-gaming applications, which makes me think maybe they'll make a push in one of those fields. I didn't want to sound like I don't want AMD to be on top here, I am happy either way. I just don't see Vega making a significant impact on NV sales in the longer run.

 

There's also 12nm LP Vega refresh to take into account (which is likely going to compete with Volta next year).
The new manuf. process will probably be used to increase performance and reduce power draw, or at least reduce power draw.
Also, we don't know what AMD might do to improve Vega further for the refresh (if they do anything at all).
We do know that Vega benefits most in performance department when HBM is overclocked for a minimal power gain (whereas a core OC produces minimal performance gains for a much higher power draw).

Plus, we noticed thus far that DX12 games seem to work better when you pair a Ryzen setup with Polaris or Vega (whereas mixing Ryzen with Nvidia doesn't produce same effects).

 

870TMG looks like same amount of ports as 870KM so i doubt theres any change at all other than mobo

 

Don't operate on the assumption that you can just swap a TMG board into a KM though, while they may appear identical it's not always possible and it's been an issue in the past.

 

im hoping is that even if it looked the same on the outside, the inside cpu heatsink will be changed for the better, MUCH better than previous 870km versoin.

 

This is what I was referring to, internal changes may be much more extensive than external.

 

They have to!! Or this will be a big mess. See power consumption between stock clocked 7700K vs. 8700K (the whole system) but the numbers should be comparable anyhow.

https://www.techpowerup.com/reviews/Intel/Core_i7_8700K/16.html


http://www.tomshardware.com/reviews/intel-coffee-lake-i7-8700k-cpu,5252-12.html

 

That's gonna get toasty even with a bigger heatsink.

 

well i have a feeling disaster is coming up. i originally wanted to grab the 8700k from SL, disable HT to reduce like 30w of extra heat under load, and run it at 5ghz in the new laptop 870TMG but now i think of it, kinda dumb and waste of money. especially with 2080 possibly be just right around the corner in half a yr.

if clevo can't be bother to add in 4th fan with a new design laptop for cpu, not worth paying premium for it either. no point putting in a 8700k to run it at stock, a K cpu at STOCK LOLOL

 

Only coffee ain't enough for an upgrade if you already have a working system.

 

Yeah, I'm a little worried about that myself.

 

Intel i7-8700K Review vs. Ryzen: Streaming, Gaming, Delidding

AMD ThreadRipper makes the Intel 8700K and AMD 1700 look like toys...with the AMD 1700 matching the Intel 8700K for $100 cheaper!

Intel Core i7-8700K Review, The New Gaming King!

Wow, the Ryzen CPU's + Vega 64LC outperforms the 8700k/7700k!!

 

Regardless of who comes out on top in any particular bm on any given day, it is telling that AMD's processor prices have come tumbling down, and fast.

With $90 to $150 off of select Ryzen cpu's it is clear retailers in my area, if not AMD themselves, know which is the best value today. Even with those price cuts, the Intel offerings are still cheaper and arguably superior for not just consumers, but for many professional workloads too.

The tables have turned swiftly and I feel that it's AMD that now needs to speed up it's timeline to keep pace with Intel. Lower prices by themselves mean nothing to me. I want more performance (from both sides), period.

As many here know; I've never felt that the performance 'crown' was taken away from Intel for my workloads by AMD's recent offerings (my current platforms (then and now) already exceed what AMD offers me). But it most certainly has changed what Intel can offer my company today (performance-wise).

The worry from the O/C'ers here (from both sides) is also (thankfully) a distant memory from me. Stability is king and sustained performance is the queen. Future productivity increases are only vs. what I am running 'now'. Not what may be possible 'theoretically' for an unguaranteed lifecycle of intermittent 'greatness'.

I like to come here to read how the 'winners' side seems to trade depending on what is tested.

Mostly though; I love the choices we now have.

May the battle rage on!

 

As I get older this is me, I'm starting to choose products that are more than I need out of the box and running them below their max spec until I need to replace them. My days of squeezing every MHz out of chips is permanently behind me at this point.

 

When a few MHz mattered (because it could make a disruptive performance change), I too chased it. Mostly because I had time and little $$$.

Today, it is easier to overbuy (i.e. an i7 QC - soon to be a HexCore w/16GB+ RAM to check emails with) than to buy something that 'fits' better, is cheaper and would do the workload as easily - but would require me to reconfigure a new platform sooner too. Time isn't just money.

Time is priceless.

A few $$$ more for an 'overbuilt' platform is easily worth it from the right perspective (age).

 

And that's also why as I get older "It just works" is an excuse for liking a particular item that I will accept every time.

 

Intel i5-8400 vs. R5 1500X Game Streaming Benchmarks

 

Delidding a $1000 CPU - Worth the RISK??

 

So, pretty sure steam is purposely trying to skew numbers for Intel CPUs. I have not used my laptop in months. I was using my TR desktop build. Never got a pop-up asking for the hardware survey. I go back to my laptop while waiting for the parts to fix my GPU block. Every time I open steam, I'm asked for the hardware survey. I had steam open the first couple days of the month on my TR build. One day later on the laptop, I am asked. This makes me have ZERO faith in the steam hardware numbers, a point of contention and debate in this thread months ago. Discuss.

@hmscott @TANWare @ole!!! @Papusan

 

intel is hard to beat now. with 2 core increase and higher frequency, it wins in most application that uses same amount of cores as an AMD chip. i think the hardest part for AMD to overcom currently is the i3, 4 cores 4 threads with an IGPU at below $180 thats hard to beat.

since AMD need market share, the largest market area is consumer desktop/mobile with just 2-4 cores and iGP. on higher end theres still plenty of reason to go AMD and future proof oneself with more cores but thats not enough. AMD cant win with just price cut it'll become how it was previously with bulldozer just a bit better off, they need to bring out APU and fast, 2018 still several months away.

could it be TR is new hardware and steam just havent catch up to it? like if steam can't recognize certain type it wont even ask, something like that.

 

It doesn't just detect your hardware and report back? I thought that was the reason I'm not getting them any more.

 

Interesting. Though more interesting that they mentioned they based the last such assessment on a delid without a tool to do it.

 

Optane 900P SSD, Successor to the Intel 750 Series-Techpowerup.com
@ole!!! @tilleroftheearth (The ssd lovers )

"Industry contacts" have told TweakTown that a successor to Intel's well received 750 series NVMe drives will launch within weeks. The second generation Optane 900P SSDs will be available in both the U.2 and AIC (PCI-E 3.0) form factors, with capacities of 280, 480, 960 and 1500 GBs. Like the Intel 750 series, the Optane 900P has enterprise roots, namely Intel's ridiculously expensive 375GB P4800X; which we covered earlier. As the 900P is aimed at consumers, maybe more accurately prosumers, prices should be "affordable".

Optane 900p

• (Rumored) Sequential read/write speeds 2.5/2 GB/s
• (Rumored) Random IOPS read/write 550,000/500,000

If accurate, the sequential write and random IOPS write speeds of the Optane 900p are substantially higher than the Intel 750 series, whose performance varied with drive capacity. It is also unconfirmed at this point, but the 900p might be compatible with AMD X399 motherboards, given that some motherboard manufactures have included a U.2 connector.

 

Any step forward is welcomed... but I want Optane in my notebook!

Almost any (i7 QC+ w/16GB+RAM) desktop that I use is quick enough today - even with just a 7200RPM HDD. (Seriously).

But a notebook still feels like it is working under molasses to me - no matter how well spec'd and current it is - vs. a properly equipped DT platform.

Intel, give us mobile Optane already! I have the cash shovel waiting!!!

 

What you have to do is convince AMD to make something like it for notebooks.

 

6 and 8 core Ryzen laptops from Asus anyone?
https://www.asus.com/uk/Laptops/ROG-Strix-GL702ZC/
Comes with an SSD 256 GB and 1TB HDD, 16GB RAM and RX 580 (4GB).

Though, other OEM's would be nice to incorporate Ryzen too and tweak it to allow for higher speeds and of course modify Vega too to work on a laptop (hey by undervolting Vega, power consumption can drop over 30% bringing it in line with Nvidia easily enough - Vega 56 can be equivalent to 1080 at lower power draw than 1080).
If OEM's can shove a 1070 and 1080 into a laptop, no reason Vega can't be either... with undervolting and minor underclocking it would probably fit inside a nice TDP (and AMD hardware can't be treated the same like Nvidia as we know it needs some undervolting because AMD always overvolts to improve yields).

But OEM's are really stupid for clinging to outdated cooling solutions in the first place... they can likely redesign it entirely with relatively cheap and far superior methods by now, but they are too lazy.

Look at laptop screens - they are usually maxing out at 1080p... and higher resolution options are rare to find, only to be ridiculously expensive.

Speed-wise and how fast a system works... meh, laptops are already on par (especially with SSD's... can't help but think that HDD's should be done away with and entirely replaced with SSD's in the first place).

 

Agree 100%, the sooner we can get rid of HDD/ODDs the better.