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

So, no longer will we have complaints of being off-topic. Here is the AMD vs Intel and AMD vs Nvidia thread as it relates to a system with the CPUs. Compare to your hearts content!!!

@Papusan ; @TANWare ; @hmscott ; @Rage Set ; @tgipier ; @triturbo and everyone else I forgot to tag. I'm posting the link in both threads.

So Nvidia and AMD GPUs can be discussed in relation to what is best for the system and which works best in relation to a CPU, but not just the direct which is better absent tying it to the system.

 

Intel Core i7-7740K Sets Kaby Lake CPU Frequency And RAM World Records-tom's hardware.com

"The new Kaby Lake-X models are very overclocking friendly when paired with liquid nitrogen and helium, but we aren't sure of overclocking ceilings with more accessible air or water cooling just yet. The chips are poised for release soon, and one would imagine that reviews are well underway, so we should know more details soon."

 

To build on that, compare the 7700K record on LHe: http://hwbot.org/submission/3412512_der8auer_cpu_frequency_core_i7_7700k_7328.3_mhz

Sent from my SM-G900P using Tapatalk

 

A little on TR pricing:
http://www.fudzilla.com/news/processors/43834-amd-entry-level-16-core-threadripper-could-cost-us-849
http://www.pcworld.com/article/3200...yzen-prices-ahead-of-threadripper-launch.html
https://www.pcgamesn.com/amd/amd-threadripper-price-rumours

Last one: "Bits and Chips go on to speculate that the highest-clocked 12-core Threadripper variant will likely be priced higher than the entry-level 16-core, giving customers a choice between the highest speed or the highest core count. Though that sounds even more wildly speculative than their pricing."

It is worth remembering the 12 core boost is 4.0 with 4.1 or 4.2 as the XFR speed. This means potentially having a 12 core all cores (with sufficient cooling) right behind intel, whose 10 core is looking at 4.3 on water.
https://www.guru3d.com/news-story/c...up-to-10-cores-first-does-4-3-ghz-on-lcs.html


And here is the company's talk yesterday: https://seekingalpha.com/article/40...bank-america-merrill-lynch-global?part=single

 

Hey, just so you guys know, an image only response isn't allowed via the site TOS, so watch yourselves.

If you are gonna build emotional tension by making a funny image response, add some text explaining what you mean by it - so there is no confusion

"You mean I can't just eat my popcorn and enjoy the show?"

 

IF that translates 1:1 to oc ability on air and water we should see kblx hitting 5.3 ghz stable vs. 5.2 on the 7700k (exceptions confirm the rule, naturally

awesome, this would give customers even more choice and close the frequency gap to barely one multi go TR baby!

Sent from my HUAWEI NXT-AL10 using Tapatalk

 

This is very interesting, X399 had more than 2x the interest by viewer count during the Computex 2017 AMD vs Intel coverage...he's got a long preamble, you could start at 4:10 given this lead in...

Remarkably Poor Performance (viewer count) of X299 Coverage (at Computex 2017)

 

i really wanna see amd destroy intel this yr, just so we can get better cpu next year from both companies. cpu hasnt improve much tbh and intel's new turbo boost max 3.0 is the only thing i hold interest in, and of course 8 cores at 5ghz would do too but thats simply uses too much voltage.

 

Keeping in topic here is a link;


Maybebe Apple but it is Vega;

 

Allot of what I have been saying;

 

Intel reacts to AMD Threadripper

 

im rolling on the floor laughing my butt off.... SO HARD to concentrate on both the audio and subs!

 

This is funny, but let's keep it concentrated on the comparison. I'm tightening the purview of it to exclude the Nvidia and AMD Vega comparison, after getting messages requesting to tighten the nature of this thread. Please see the OP in about five minutes for the updated rules for this thread. But this was funny...

 

Control freak much?

Is this a conversation/discussion or an accredited course (how many high school credits do we get for completion)?

You do know that 'the answer(s) you get will depend on the questions you ask'... we already know how this is going to go down (assuming Intel and AMD deliver what they've promised so far).

Good luck with your preaching, er, I mean topic.

 

I would say this video mentioning the comparison Aka on-topic

 

I hear ya. It is a good summary of Intel recently.

I had to put more rules on, etc. After being asked to narrow focus of the thread, hence trying to help out with that. That's why, although I loved it, I'm trying to find the exact lengths of how far to prevent flame wars and keep it on comparisons.

Sent from my SM-G900P using Tapatalk

 

Wow, you had a great idea, and someone thought they would cap it at the head and ruin it for us.

Whoever that was, please stop trying to change things into what you want, and please let us make it what we want, ok?

I was looking forward to participating in this thread for deciding what to get moving forward from the wide range of AMD / Intel competing products newly coming out.

IDK if I want to pair Ryzen with Vega or Volta, or jump to Threadripper to get full performance from Vega / Volta.

Or, will AMD drop the price of Epyc so far into the realm of my potential interest that I may try to build a system using it's vast headroom - start small and have a platform to build it up.

And, via some miracle, if Intel gets their mess put in order and drops prices to compete, will there be some instances where I can use Intel as a solution - not for me, but for clients. It's possible

Restricting the topics down to what someone else wants to focus on isn't what @ajc9988 and the rest of us signed on for in here.

Please re-open up the thread to what it was so we can have the freedom to make it work for what we all need, room to discuss the options available moving forward for building systems with Intel and AMD.

Thank you.

 

I hope the <very> long rule list in Op don't hamper the discussion

 

@ajc9988 who are the ones trying to mess up the thread by changing it into something completely different?

What's up with all the restrictive rule making all of a sudden? This thread was starting to get interesting...now, not so much.

Maybe they should open their own thread and set it up like they want it instead of taking over this one?

Howz about we leave this one and go try again with a new thread?

 

The whole op is a rule Box. No need for it. Grown people should manage to have a normal discussion, without a flame war.

 

So, no one will be called out or named. It was concerns over flame wars, which are against forum rules, and that people might be afraid to post due to getting lost in the thread and buried with how wide the content is. It isn't by force! I decided, to deal with these concerns, to try to limit it in a way where we still get what we want without the most contentious elements likely to lead to all out brawls.

Edit: I can change the rules. It is no problem. I just wanted to give a list of the reasons why. Price per performance is in the table, as is anything that is used by the cpu/chipset, meaning which videocard works best with what CPU. For the most part, this thread has been tame, but I can understand concerns on all sides.

Sent from my SM-G900P using Tapatalk

 

No need for reply back on a post if someone is trying to fire up. + The rules on the forum should be enough to prevent it.

 

Well, the people seem to have spoken. I'll remove it in a moment.

Sent from my SM-G900P using Tapatalk

 

For the record; I've never seen a 'flame war' here.

That is why I'm here; people are more mature than on other forums I've run from.

 

And the way we like

 

Well, maybe there needs to be different threads focused on what others want - they can narrow it as they like to make their threads what *they* want.

There are no flame wars here, why overlay some made up FUD to control how this thread is run? If anything what they are doing is creating friction, we were doing just fine.

Wait till problems happen before over-thinking things and we are all sitting here with nothing to talk about

We branched out from an AMD Ryzen thread to have an open thread to *increase* the breadth of discussion as the limits of the other thread weren't working, to get away from those strict limits.

We were being told multiple times a week what we can't talk about and then the same people start going off onto tangents of their own. Well, they've got that thread to play in, why can't we have our thread to play in?

I don't understand why there can't be focused threads like they want, and an open thread like we want.

@ajc9988 what do you want? What was your original intention with starting this thread?

 

It is resolved. As long as we stay with comparisons. I miss my analysis of the Intel process changes, which would contrast great with AMD's new disclosed processes moving forward, which is right on topic here.

I think it is a concern on manageability, but I do plan on checking this thread regularly (even if I don't respond immediately). But, let's see where this goes.

Edit:

To have a place to openly discuss and compare the two companies offerings (Intel and AMD). Because AMD also has graphics, and that is highly awaited and related, having that, with the arguments surrounding whether they waited too long and Volta should be the focus or not. In both Intel and AMD threads, we went from Ryzen to TR, or talking about the new HEDT offerings from Intel. Because those directly affect the server offerings, and the HEDT now have the core offerings of server chips, it only makes sense to discuss them directly with the HEDT offerings. Then, if you are talking about the chips, that means nothing without the motherboard and chipset to support it, which also leads to ram with how it effects the infinity fabric speeds. So, it is a narrow universe, but there were concerns. This seemed like the best place to hash out these differences.

 

That's too bad, this thread just became uninteresting to me.

Without the whole system open to discussion, taking away any discussions integrating the video cards in particular, what's the point?

 

Reposting my edit you may have missed:

To have a place to openly discuss and compare the two companies offerings (Intel and AMD). Because AMD also has graphics, and that is highly awaited and related, having that, with the arguments surrounding whether they waited too long and Volta should be the focus or not. In both Intel and AMD threads, we went from Ryzen to TR, or talking about the new HEDT offerings from Intel. Because those directly affect the server offerings, and the HEDT now have the core offerings of server chips, it only makes sense to discuss them directly with the HEDT offerings. Then, if you are talking about the chips, that means nothing without the motherboard and chipset to support it, which also leads to ram with how it effects the infinity fabric speeds. So, it is a narrow universe, but there were concerns. This seemed like the best place to hash out these differences.

Now, peripherals are on topic, except for the Optane vs NVMe argument. That one drives me nuts. Mentioning Intel can use it, fine, but beyond that...

Videocards are another thing, especially with the tests showing it works better with certain chips. That would still be on topic. Also, the OP has already been changed back.

 

Maybe add the Vega / Volta back into the Title lest someone claim somewhere down the road that discussing the Video cards in relation to the CPU's isn't On Topic?

That way someone can still have another thread just talking about the video cards individually while we can discuss the whole system interaction here?

 

AMD 16c at 4ghz
Intel 12c at 4.3ghz
intel has an IPC edge over ryzen by about 6%

going by 100% scaling, AMD will be at 16 * 4 = 64, Intel be at 12 * 4.3 * 1.06 = 54.7. now we all know 100% scaling is impossible, the actual fact more cores means scaling is poorer as per ANANDTECH and their TSX article. these CPU do support TSX however software side of things will also have to support TSX which consumers software wont have them prob for at least another 7-8 years LOL.

so say best case scenario we take off 5% performance off AMD's chip due to 4 more cores and crappier scaling than intel's, 64 * .95 = 60.8.

basically for 4 more cores with AMD, we get about 10% faster in multi threaded scenarios, however in single / dual core cases, intel's turbo boost max 3.0 can probably ensure two cores running at 5ghz thus beating AMD by quite a bit.

for price comparison we got $850 vs $1250 ish oh and intel using toothpaste under IHS

 

Doing it by math is far too complex for most. In essence if the app uses just 1-4 threads the Intel CPU's usually win. If however they are using more than that or even if you are running multiple CPU resource hungry applications the Amd's Ryzen is the overall winner. The caveat to that is if just a casual user browsing the web, doing lite office work, managing media files and the like you may never notice the difference between AMD or Intel.

That is just a general rule of thumb. Of course it is more complex than that. What will make things a bit muckier is x299 competing against the lower Ryzen cores. That is Ryzen 3, 5, and 7. The quad channel memory (where available) and other enhancements may change things a bit there performance wise on the x299. Don't forget on the lower end the stranglehold on PCI-e lanes too.

 

Link;
https://www.fool.com/investing/2017/06/11/in-another-blow-to-nvidia-amd-sets-its-sights-on-t.aspx

 

Also, you give AMD a huge hit when their multithreading, I believe, scales better than Intel's. That is where, I believe, we should take issue. Also, I haven't played with boost, but I would be dubious at claiming 5ghz on 2 cores. If you are able to independently set the boost, even then, this is skylake. 4.3 is the max for the 10 core, not the twelve core. Assuming it is like any other chip, the two extra cores drop the cpu by 100-200mhz, meaning your analysis is off on speed as well.

Now, if you can show me on boost 3.0 allowing independent OC of those two cores, I may have to restate. But you have a lot of assumptions in there. Also, could you post the TSX article and has anyone tested if it is TSX specific for that scaling?

Edit: also, with TSX, it took Intel 3 years (from 2012 when introduced to working, finally, in 2015)...

Sent from my SM-G900P using Tapatalk

 

The 10 core is not going to compete against the 16 core TR, there is just no way under load. It does seem you can get 4.5 on the OC and delided one guy is getting 5 with water. Now 4.5 would trash a Ryzen 1800x but at the cost as well. It would be a pretty good stop gap between A Ryzen 1800x and a full 16c,32t TR. A 12 core Intel would be a better stopgap though. Again at a possible higher cost.Not knowing the cost of a x399 and x299 along with the fact you have to have high speed ram for the AMD but could get away with less for the Intel. it is hard to guess the end costs.

With no info on TR I am highly leaning toward the Intel offerings at the moment. This again is AMD's fault, so while there may not be a 18 core for a while a 10 core may suffice for a year or two.

 

I still feel that the wait is best, as it gives a couple months to work out the kinks. Also, I'd want to know, considering his is binned, what the average is. This is with 10 core and every 2 cores will drop the AVG OC, as it always has. Still impressive. Also, Intel will be supporting the G.Skill 4800 DDR4 (board dependent), so we still need to wait to see what AMD will support.

https://gskill.com/en/press/view/g-...017-with-new-intel®-core™-x-series-processors

 

I do not think with quad channel ram Intel would be as in need of DDR4 4800 as much as AMD will be. Intel does seem to get better with high speed ram but not much so as Ryzen cores (could well be wrong here). Not really willing to wait a few months here, I doubt I will last the month.

TBH with TR it feels like with the info we have been getting, and have, we are looking at Christmas season shopping. This is the same for Vega, if not worse. Again this is all AMD's own fault, they are holding information too close to the chest. Because of this they will loose a few early adopters and prior fans. I am sure they are fine with that.

 

I agree AMD would benefit from higher clocked ram, but wanted to share info on that platform. Now, I do wonder if all of the ram speed being lower is the IMC or if it is based, in part, on the hearing of infinity fabric (in other words, whether that causes something effecting stability). Complete unfounded conjecture on that.

I can understand how you feel, just Intel hasn't given me anything to jump in this early. I suspect they plan to drop rumors and info in the next week right before Intel i9 and X299 go on sale, that way you take wind out of their sails on launch day (dick move, but fair play). If you have to launch a month later, but your Epyc launch is within days of Intel's HEDT, you plan on taking details public then if you have a card up your sleeve, especially since Epyc will let the cat out of the bag on some of this anyways.


Sent from my SM-G900P using Tapatalk

 

i aint playing favorites when comes to the company cpu, i like amd more as a company but i like intel's cpu more because they are faster. i simply did a quick and simple comparison and try to include some estimation thats it.

when you mention AMD's cpu scales better in multi threaded workload, can you explain why?

 

Yeah, you aren't doing anyone any favors by pre-judging Intel as "faster", as it depends on the work load and the variance is small enough even then to not be important even for long jobs.

More cores are also "faster" in a lot of work, so in that case the same $ Intel CPU isn't at all "faster" than an AMD CPU for the same cost, and in this case the time saved by an AMD CPU *is important* for long jobs

 

Simple. BW-E has about the same IPC as Ryzen 7 1800X. That is not disputed. But, look at the three following scores in Cinebench:
http://hwbot.org/submission/3254480_meankeys_cinebench___r15_core_i7_6900k_1969_cb/
http://hwbot.org/submission/3515023_ace123_cinebench___r15_ryzen_7_1800x_1951_cb
http://hwbot.org/submission/3532112_synsbg_cinebench___r15_core_i7_6900k_1934_cb/

So, in other words, it takes the 6900K running at 4.6GHz (4.7 to get 1969, 4.5 to get 1934) to equal a Ryzen 7 at 4.2GHz on multithreaded applications. This has been seen in varying degrees in other scores as well. Now, obviously Skylake-X will have higher IPC, closing that variation to a degree, but the only data point we have on that is CPU-Z multithreaded test, showing a 3% improvement over BW-E. This means I don't hold out lots of hope on improvement this generation on Intel's side. This is before seeing any improvements or performance hits of TR, so take with a grain of salt at this moment, but keep in mind that Intel is not as impressive in some tasks, although they rock at lightly multi-threaded and high IPC favoring workloads.

Note - I have not controlled for ram, MB, etc. in this quick answer. I usually do more to control for variables in the build.

Edit:
http://wccftech.com/intel-core-x-core-i9-7900x-vs-core-i7-6950x-cpu-benchmarks/
5198/5030=1.033399602385686 or 3.34% improvement in multithreaded test.

 

My guess on this one is because there are no tricks, including NUMA that are played in the scheduler. Since every thread tend to be more evenly spread out between all cores the workload is more balanced between the cores. This right up to the point where the CPU becomes fully loaded. Like with my current 4c,8t I can watch the boucning at times all over the 8 threads.

 

30 day review, after 45+ days... AMD 1800x vs Intel 5960X

Ryzen 30 Day Challenge Result: I'M NOT GOING BACK TO INTEL!

 

Only 6525 on a dual CPU system, this is to compete against a Dual Epyc, I think AMD stands a chance too really bury Intel here. Come on AMD just give it up already!!!

 

no, intel's cpu is faster, going down to per core and it's ipc, nothing to argue there, no one disputes that.

yea i thought I'd see cinebench results here. we also got to know, ryzen's latency within the same CCX is about half of intel's as shown by pcper, however when it uses infinity fabric and workload exchanges between CCX the latency is easily 3x higher.

cinebench and handbrake does NOT shuffle workload/data between cores/threads, not simply because when benching all cores/threads are at 100%. thats why when using cinebench/handbrake they do so well in those results. however real world software aint like that, not optimized for ryzen

 

That doesn't make Intel a better choice at all, it's only 1 element to consider when buying a CPU.

The cost and the core count, real work throughput, count for more.

A per core IPC advantage isn't enough alone to make a decision.

 

So, first, NO ONE has retested the early results on Ryzen on many professional software since improvements have been made at the microcode, bios, os, and software levels. Further, what I am posting from Tom's Hardware was before the ram was unlocked, before it was known of the 20 degree offset, before the ram gearing was understood, and ran at 200 mhz lower than the consensus overclocking level of the chip (I really don't like some of their bias there). With that said, I'd also like to mention the POV Ray change right as Ryzen came out which gave Intel a significant bonus, while simultaneously hitting a bit on Ryzen. But, supposedly it was a scheduled change. Anyways:
http://www.tomshardware.com/reviews/amd-ryzen-7-1800x-cpu,4951-4.html (March 2 testing, meaning likely done before release, when things were real rough)

Considering 3200 is being reached by many more people (although I see them using 3200 kits, not higher, so haven't seen people try for or achieve much higher than 3766 at the top end of ram speed), these tests should be re-run with the mods by the software companies, especially considering 15+% has been normal in most applications since release.

 

dude, you quote me stated that i "claim" intel is faster which it isnt, now i explain to you why its faster then you make claim i didn't talk about "value". rofl, orginal post i simply mentioned intel is a faster cpu, period, which it is regardless of value. no $h!t Sherlock AMD has better value duh.

if we dont talk about value theres other cpu too, like ibm's power 8 and power 9. 8 threads per core clock at 4.5ghz, 12 cores all at 4.5ghz 600w tdp rofl

 

theres really nothing to retest on cinebench and handbrake, those software has not update at all, it is already coded so that it benefits AMD's cpu more, why do you think AMD would run it to bench and show off their scores against intel? it should be bloody obvious lol.

on the other hand, look at civ and AOTS which both games are heavy multi threaded yet AMD was getting destroyed until game developers made update, which is still behind intel atm. imho i think its all software now, AMDs done what they can, its time for software to catch up. and truth to be told, intel side is the same, they coming with new extension and software developer not taking advantage of it except in the enterprise market. AVX was out since sandy/ivy iirc thats over 6-7 yrs ago, yet out of like 100 software i use only select few that take advantage of AVX.

software and hardware goes together.

 

This is what you said:

And, I responded that Intel aren't faster in all measurements or considerations, and to flat out state that Intel is faster is wrong.

You keep coming back saying Intel is faster, and I keep coming back and telling you AMD is faster and cheaper, when taking real work all under consideration -> Intel isn't faster. AMD is faster.