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

The more I see, the more I believe that this is a really bad time to buy into new tech. Better to kick back, grab some popcorn and see what happens. Spending money on new AMD or Intel stuff right now would probably not be very smart. I doubt that anyone here really knows what might happen next, but speculation is fun for some. Some enjoy playing ping pong. Some prefer Thunderdome.

 

I agree, especially with DDR5 supposed to be fully finalized and published next year (summer 2018); which Intel implemented DDR4 on board revisions to support it, so may do the same with the X299 platform, or just integrate it into Ice Lake platform. Zen 2 is the same, with the 7nm process supporting both DDR4/5. Now, if you want or need to buy now, the Zen socket lasts for Zen 2 & 3. Intel's doesn't. Meanwhile, if you buy now, you are probably doing so because you want to upgrade once PCIe 4.0 is ready around 2020. It takes a couple years to get really good, high clock rates on ram anyways once the standard is published.

Me, I planned on HEDT, Intel let me down, and AMD has promise (and a socket wired to handle many more cores than given, as the Tom's Hardware article pointed out there is no difference between the EPYC socket and the TR socket).

 

Well, we still need some meaningful action on the GPU side of things, and there's really nothing reliable available to know if there is actually anything we can count on looking forward to. We need some serious scrapping to be happening. Scratching, kicking, screaming, biting, punching below the belt... all that's fine. I'm just tired of the GPU war being a single player game with one gladiator and a waterboy.


Hopefully, they won't both end up going down the namby-pamby treehugger road like what has happened with all the castrated BGA laptop filth. I'm already seeing too much crap talk about using less power. I'm like who give a rat's ass about that, show me the horsepower. I want to see devastation and destruction, and I don't care if it uses more or less power to get the job done. I will say this much... when I saw how massive Threadripper CPU was my heart skipped a beat with joy. I'm like yeah! Bigger! Right on! I think bringing Tomahawk missiles to a knife fight is aways the best policy. You win and you don't even have to get your hands dirty.

 

yep the power of software optimization is scary. which is why i put in a lot of my time to hunt software and their features.

 

TANWare, I'm trying to find the original post of mine that Mr. Fox quoted in post 1086 and there have been many replies to (or about).

Was it deleted or moved?

 

Yes it was deleted as it was part of the back and fourth bickering of you two, I am not playing any favorites and no edits to any posts but the response of 1086 was relative to the thread and not part of the bickering. I do not think others should suffer from you two!

 

Bickering? Man, I go away for just a little while and miss all the action? Dog gone it!

 

wot in ternation happened around 1086?

 

To fill you guys in, tiller and I have had back and forth battles a lot, much of which has had to be cleaned by TANWare. Because of this, we have both been warned with being banned. What was removed was a couple comments, which was to stop it before it started again.

Sent from my SM-G900P using Tapatalk

 

About intel v amd?

 

The problem is not that it was Red vs. Blue, the problem is it all was the "you are not listening to me" and getting personal. All off topic with some minor points here and there. one or two sentences in paragraphs of rants at one another. Also when I say stop it this does not mean carry on for another several posts each.

Nor does this mean taking shots off topic now either!

 

deep down, its blue vs red

 

I think you mean red vs blue (roosterteeth reference; great show)!!!

Sent from my SM-G900P using Tapatalk

 

Ryzen 7 1700 vs. Core i7-7820X, 8-Core Royal Rumble!

Huge Power Draw from the i7-7820x, almost 100w more than the 1700 when both OC'd!

 

Look at this beauty.

 

To all the naysayers. ASRock have confirmed that Covfefe Lake needs a new chipset.
https://linustechtips.com/main/topi...be-compatible-with-z270-chipset-motherboards/

 

Let's give some credit to the source too

Tech #CuttingEdge
AUG 2, 2017 @ 07:38 AM 1,190 12 Stocks to Buy Now
Intel's Coffee Lake CPUs Not Compatible With Z270 Motherboards Says ASRock
https://www.forbes.com/sites/antony...h-z270-motherboards-says-asrock/#33dbecb76a2d

"PC enthusiasts are waiting for several exciting products to hit retailer shelves in the coming months, but one of the most anticipated - Intel's Coffee Lake six-core CPUs, which are the company's successors to its current Skylake CPUs, will not be compatible with current motherboards."

Intel's gonna knock it up another Notch!!

Spoiler: Credit where credit is due...

 

its really sad. look at intel trying screw people over on desktop side. compare their mobile and desktop platform, http://cdn.wccftech.com/wp-content/...ee-Lake-and-300-Series-Platform-Details_3.jpg

look at all the additional features come with mobile side of things rofl, while desktop want people to push for yet another socket. had not being the highest clocker and performance CPU i'd have gone AMD. if only AMD could do 4.5ghz.. sigh

 

Well, looking at the leaks regarding the TIM (same crap paste as Kaby Lake) and the 50% increase in core count of the exact same design would also result in a 40-50% increase in power draw and heat. I'd say Intel won't be going far past 4.8GHz this time even with a Delid. Without a delid, I'd be amazed if these CPUs didn't have issues out of the box. Oh, and to make matters worse, the first wave of boards won't have any new features at all. The boards with the additional features are coming Q1 2018.

Sources: LinusTechTips news forum (too many topics to list)

 

its actually the method intel has been doing for over last couple of years. treat enthusiast desktop consumers like trash and give mobile new features, while pushing it and give crappier BGA silicon.

only this year, because of AMD, intel pushed desktop segment forward other wise just like all previous years the mobile platform would come prior to desktop which is really sad. intel is trying hard to control their clock speed and they know they have got nothing to improve except for adding features or clock speed rather than improving IPC. if not for AMD they would not blatantly advertise their clock speed and offer a 6c at 4ghz turbo, more like 3ghz 3.4ghz turbo with 6c.

when ryzen2 hits, as soon as theres at least 4-5% IPC improvement ontop of that with 500mhz higher overclock then its time to jump to ryzen. assuming temperature is good and all.

 

This is why a 1950X will get me by until Zen 2. Even tech sites are excited for it (except pcper looked underwhelmed and tried to minimize it).

Sent from my SM-G900P using Tapatalk

 

even though right now we wont know if ryzen and ryzen 2 will be same socket. however knowing its AMD, 95% that it will be same socket, just because its AMD rofl.

if ryzen can implement something similar to intel's TBM 3.0 boost 1 or 2 cores to like 4.6-4.7ghz then it's worth going ryzen.

 

Actually, we do. AMD announced a while ago that Zen, Zen 2 and Zen 3 will all be AM4. AM4(+) will be supported actively until 2020 at least.

Additionally, we also know that Zen 2 is pushing 5 cores per CCX so the next generation mainstream CPU on AM4 will be 10 cores with improved clockspeeds and then going down I imagine at 8, 6 and 4. As for the next TR4 CPUs - likely starting at 20 cores.

 

First, AMD already announced the same socket will be used through the next couple generations, so Zen 2 and 3 is likely.

As for the boost, it is a 3.4 base/3.6 all core boost, 4.0 four core boost/ 4.2 XFR (one or two cores). I can't wait to see the regular overclock, which so far is 3.9-4.0GHz, potentially.

Sent from my SM-G900P using Tapatalk

 

Do you have an article on the five core ccx?

Sent from my SM-G900P using Tapatalk

 

It was posted somewhere in this thread IIRC - or at least I'm certain that it was in one of the NBR threads about Ryzen.

 

5c ccx thats a rather odd number. maybe two 8c ccx disabled 3cores each to 5+5 for 10c sounds more realistic.

i know XFR is there but 4.2ghz is too low. also we have know idea how the 1 core 4.2ghz XFR works btw. im guessing its more similar to turbo boost 2.0 rather than TBM 3.0.

 

XFR is all about ease-of-use. Assuming you have good cooling, the CPU handles itself independently and boosts where necessary.

I'm sorry but I don't agree. We've already seen that Intel's new Mesh design has drastically reduced IPC, to the point where a 4.5GHz 7800X is the same speed as a 4GHz R5 1600.

 

about your mentioning on XFR, that sounds like turbo boost 2.0 yeah?

regards to your reduced IPC, in games yeah?

 

In everything. Check HardwareUnboxed - they tested 30 games and I believe a few synthetics. The 7800X came out barely ahead for 130% more money. That's horrible value. Not to mention it ran much hotter and drew a LOT more power.

 

i dont see any synthetic benchmark for ST performance when i search hardware unboxed videos, i only find benching games. i do see other places for CB IPC for skylake-X and kabylake is the same. so basically regards to reduced IPC, just in games right? due to un-optimized for the hardware

 

You see ipc for one task: cinebench 15.

Sent from my SM-G900P using Tapatalk

 

cinebench is a well optimized software, games and games engines not as much.
also i found this http://www.eurogamer.net/articles/d...view-core-i9-7900x-i7-7820x-i7-7800x-i7-7740x

somehow it seems hardware unboxed have really messed up results vs this site.

 

Nope - HU haven't messed up anything. The results have been validated multiple times from what I've seen. The newer Mesh design introduces absurd latency in Intel's designs.

 

it does add latency, i wouldnt call them absurd. versus the eurogamer review site, i wouldnt be surprise if we find out hardware unboxed review cpu throttled lulz

 

They used a 240mm high end AIO on the 7800X - I mean, if it had throttling issues, that's Intel's fault for the CPU not being able to hold its base clock at stock settings.

 

theres plenty of throttle type, cooling is one of them and did not perhaps not see der8auer's video about what type of throttling are there. anyway, its a guess from me, if one would want to explain the different results between hardwareunbox and eurogamer then we'd have to think else where.

 

VRM throttling occurs only when you OC these CPUs - that would not explain the piss poor performance HU and many users are noting in gaming at stock settings

 

So your argument is cpu throttle after going to 4.7ghz and a custom loop cooler?

Sent from my SM-G900P using Tapatalk

Edit: clarifying position, not being judgmental nor saying you are correct or incorrect.

 

honestly no idea. in their review it says 4.7ghz for skylake-x and power consumption, the consumption could be a peak of all average peaks. theres not enough info to know if its really throttling or not but i do know that there are many type of throttling as we all know. just with throttlestop theres 3 types, clock modulation, chipset modulation, thermal. theres also current which is related to clock modulation as well, VRM temp, and many more.

also, this assumption of throttling is compared to results from eurogamer because eurogamer shows much smaller difference.

 

I'll try to pull up the numbers and settings in a side by side later tonight or tomorrow to point out hardware differences (including cooling potential of the setups) and other things I can notice to reconcile the scores as much as can be done. This will at least narrow the potential differences in the scores to get an idea (just like it varies between anyone else). Thank you to bringing the disparity to our attention so we (and lurkers) can dig in and try to get the best information to compare on.

 

R7 1700 & i7-7700K for 144Hz Gaming at 1440p, 1080p, 4K

 

I still feel like overclocking Ryzen, should also involve RAM timings, as they are a major part of the infinity fabric which yields a massive performance boost if done correctly.

 

The 7700K is the undisputed gaming King. If this is all you will ever be doing with the system you should not look at anything else.

 

4c is a joke lmao. i have so many stuff i use that would run that cpu and my system down to a halt because 4c is just too slow.

honestly once you have had 8c and overclocked, its really hard to go back, even 6c 5ghz seems barely enough.

 

wow waited forever finally E-ATX x299. but damn it all no onboard VGA. might have to wait till it hit newegg to really see the full spec. also doesn't really say the pcie lanes split for 28 lanes.. more waiting.. on info, on vroc, on siliconlottery...

 

coffeelake 6c 14nm++? https://techreport.com/r.x/2017_03_30_Intel_manufacturing_day/14nmcharacteristic.png

the picture seems too much on the positive side. going by the trend from 6700k to 7700k average chip was able to get +200mhz near the high end (4.8ghz to 5ghz) for similar voltage/power/temp, i'd give hope to see that 8000 series 4c maybe able to get at least +100mhz on the highest end (5.2 or 5.3ghz). not gonna wish for another 200mhz+ as we all know that seems kinda unrealistic given physics and heat needs to dissipate it.

8700k 6c 5ghz seems very doable.

 

You're daydreaming - clockspeed doesn't scale linearly as processes improve. Example 1 - Going from Sandy to Ivy average OCs dropped from 5GHz to 4.6GHz. Going from Haswell to Skylake you had no change in average OC. The chances of you getting a +300MHz average OC over Kaby is non-existent in this case. Remember, it took them 3 refreshes of 14nm to get 5GHz initially on 4 cores (Broadwell was 14nm, Skylake was 14nm+ and Kaby is 14nm++)

 

no im not brother @don_svetlio imho i think there are a few enthusiast share similar views. though we have previously already came to an agreement that the group i belong to is a puny tiny section.

also from what you mentioned, i did address that 14 to 14nm+ we saw an average 200mhz bump with similar voltage, going by the trend we'd see another 200mhz+ going from 14nm+ to 14nm++ but instead of 200 i said 100mhz instead because i didn't want to give too much hope. thats an intel marketing picture and they bs all the time.

also i thought kabylake is 14nm++ but @ajc9988 would like to disagree. lets call skylake 14nm .5+ instead ha


edit: @ajc9988 now i think it over abit, other than throttling issue, another possibility for hardwareunboxed video to show their results would be the resolution. though they mention in video every game is at 1080p, if there were at 1440p then it would make more sense. again donno if thats the case might have to ask the reviewer himself.

remember back then when ryzen review first came out? at 1440p the difference is barely any, but at 1080p the difference in fps was pretty big. could be that, GPU bound not cpu bound.

anything thats more GPU bound, 1080p with ultra settings everything on highest with shadows, AA etc would be similar to high settings on 1440p? not sure.

 

We already are on 14nm++ - first gen 14nm was Broadwell and from Broadwell to Skylake there were no gains to be had.