AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

I think Intel, like the rest of the world, was expecting AMD to be forthcoming with information and was ready too release. They just lost me as a potential consumer, I am sure quite a few others as well.

 

TANWare, if I was just buying a personal platform this spring/summer; AMD would have lost me today too.

They came out with all twelve pistons firing but now seem to be slipping into old AMD mode (over promise, under deliver). Hope to see that reversed (and soon).

I have no doubt that most of the boasting and puffing will be shown to be smoke and mirrors in due time (by both Intel and AMD, to a certain extent) - but I don't want to see this show stop when it was just starting to sizzle.

As a corporate user? I'd have cancelled the PO at 10:54PM EDT and would be calling my friendly as a shark Intel rep promptly at 8AM tomorrow.

 

They've been saying summer for the past couple weeks/month, and expected for August. Intel has up to 10 core this month, 12 core in August, and the rest 6 months after.

All this crap on Intel not reacting is BS. If you look at minimum tape out to sell, it is 9-12 months. That means either the horizon event to the Ryzen release is enough time to tape out if they had the rough design down, which they did. The platform has been set either way. The only question was when they decided to react. I'm betting December. Up to ten or 12 core was planned longer and is obvious by release date. All using the same platform, so it isn't the full cycle needed to get those chips (but full tape out is needed, which Intel controls the fabs, so can rush a little, if needed).

They can puts me off all they want, I'm still about the best performance for my needs. With that said, don't know I'll give them much more time after this on Vega though. That was about it on the chance to tease there. But the CPU, I'll wait and see because I'm not going to rush out for a new platform for a 10-core with a premium.

But, I can understand the frustration.

Sent from my SM-G900P using Tapatalk

Edit: remember rumors of TR started in March. That means Intel's 14-18 core offerings are about 9 months from the rumor (and the rumors inside the industry could have started earlier considering when TR tapped out and ES was available).

 

Of course Intel is not reacting. Sure; it's acting, but not any differently than any other company would (i.e. see NVidia and AMD...).

But to imply that Intel just keeps a few 'rough designs' up it's sleeve is preposterous. Processors don't build themselves and when they're actually designed/built and delivered; billions of dollars has been spent up front and for a very long lead time.

I think highly of Intel for many reasons. But being able to pull an 18C36T chip out of it's butt and promising it in roughly 8 months time is something even Intel can't do (even if it was stupid enough to devote the resources and man hours to try).

Let's be realistic here and call a spade a spade. AMD jumped first. Right now, it looks like Intel might jump farther.

 

I just have no patience for a company that is just holding onto information. release date, prices and product benchmark comparisons.

Intel may not be pulling from anywhere other than they had over developed but before had no real consumer market but it seems now they do. These higher end core CPU's may have been originally Xeon's but are now Consumer but what does it matter. AMD tossed them the ball and now they are responding. Hopefully with not as much vaporware as AMD is having as of late.

 

TANWare, that is my hope too and why I want AMD to succeed as much as possible; I don't want Intel to have a reason to back down on any of it's teases, promises, etc. either.

What would be great is to see AMD come out with a TR that clobbers current Intel offerings on more than once aspect (performance/price/efficiency/etc.)...

That will stir the giant into full tilt production for sure (with a win for all)...

 

Wait a second... rumors come out on Threadripper first (shortly after Ryzen is released) and then later on AMD confirms it, followed up by Intel that suddenly also has a comparable product for release much later.

There is 0 pricing information on either products from Intel or AMD, and Intel is likely to be more expensive (though we don't know for certain).

So, why side with Intel on this one when they didn't bother to announce their i9 any sooner and are probably only releasing it in the first place due to Threadripper?
I wouldn't go with Intel even if you paid me (and you'd be able to buy several Ryzen systems with their price tags easily enough).

As for Vega... delay was known for a while now due to HBM2 limited availability. Its not great, but at the very least we know consumer Vega is slated for a slightly later release, and AMD said they wanted to do it right... it might seem they are also gearing up for Volta so that Vega is ready for it ( at the very least they might be doing something good with the extra time).

TANWare - I don't get why you skimped out on the Ryzen system just because of Threadripper and are now willing to buy an i9 even though it looks it will be released later (or possibly at the same time as Threadripper)?
I can understand waiting for a CPU with much more cores, but I think you simply might be too impatient at the moment.
As for the GPU... you can always get a cheaper Nvidia or even AMD option and then replace it with Vega if you want to.

Maybe I'm not just getting it, but could you illuminate your thought process on this?

 

Actually, Intel did launch pricing. That and core count is what they confirmed. No show of performance. Intel releases the 6-10 core chips this month.

With Vega, timeline hasn't changed. Q3 for the consumer release. But, sitting over the Ti with little proof of worth is painful and upsetting.

Intel hasn't shown off any performance except the Kaby-X and said that no chips will be soldered moving forward. So, I have zero reason to think skylake-X is any good.

But we do want more info.

Sent from my SM-G900P using Tapatalk

 

Completed session from AMD Youtube Channel
AMD Computex Livestream 2017

 

Ah. In that case, I'd chalk it up to Intel already having a preexisting platform ( for servers mainly) that was more or less ready to be adapted for the consumer market, whereas Ryzen was not. Hence the lack of info on Threadripper (plus its a smaller company with fewer resources).

I would imagine that Threadripper will be released relatively soon (in the comparable time frame as i9), so as of right now, there's no point in jumping on either CPU before more info is released on AMD end due to a bit of 'disappointment' and a bit more waiting

Vega I can understand, but Threadripper is still extremely fresh

 

Intel Core X Series vs. AMD Ryzen Threadripper: The Details So Far!

 

Actually, Epyc launches at the end of the month. It is a server chip and probably why they wanted the gap, so they can focus on helping server clients to purchase and deploy the server chips.

Interestingly, Intel pushed the release of the i9 to August at CES. They then moved the release back to June when the rumor was that TR would be summer of 2017, meaning likely August release. Considering that the 12-core will still not be ready by June, that was what they were waiting for release until August, but with real competition, they wanted to get as many onto the platform as possible before an alternative was available. So, I'm not nearly as impressed.

It also must be noted that Intel split the socket to be used on Xeon chips this year. It no longer is the same socket as is found with the HEDT lineup. So, to design the 14-18 core parts, it would have been drawing on prior experience with the Xeon line to scale up core count. But, once again, there wasn't a reason until TR. Now, depending on when they heard, as I said, pulling it since it is stacking cores in a line and running the ring bus, I'd say it was fully reactionary to do those. Intel has traditionally only allowed the top two-three chips to have 40+ - lanes of PCIe, with one 6-core having much less. That would be the 10 and 12 core, with bringing the 8 and 6 core down to compete more directly with Ryzen. But those were planned. I'm saying the 14, 16, and 18 core chips were reaction and the timeline shows it is possible to be pure reaction to news of another product.

But, staggering the release also keeps the MB manufacturers from being dicks and focusing on Intel like they did on Ryzen. They chose to focus on Z270 over the X370 boards, then did not expect the demand that ensued, resulting in a long drought of MBs for Ryzen. Plus, Ryzen has outperformed on expected demand in other ways as well, but has had many other growing pains. By having the X299 boards released now, there is more time to focus on the X399 chipset and boards, meaning it should be more solid at release.

With Vega, you have the aftermarket, third party 1080 Ti designs now coming out with performance being shown and price stratification starting to pick up. This is why making wait on Vega with no info other than the promise of much higher minimum frame rate performance is slightly angering. There is a true market alternative already available. It's got a decent price. So, waiting is more painful than with the other, which has no market equivalent until August at earliest, with the rest arriving end of Q4 to H1 2018.

Now, Intel did not confirm anything on Coffeelake that I have read yet. Just that they are going to paste on the big chips (HEDT), which is to cause ****ty thermals, increase the number of delids, then payout less on warranties. That is a horrible business move intended to stack coffers by screwing over consumers.

 

Am I missing something, or the information is that the pricing is subject to change?

Just different interposer is my guess.

 

http://forum.notebookreview.com/threads/amds-ryzen-cpus-formerly-zen.799348/page-166#post-10536189

Obviously Intel could change pricing, but they rarely do once announced. By putting it out there, AMD could have announce their pricing and product stratification. This would have allowed for a better comparison and held off those going for it this month. Other than that, I still have my timeline and AMD hasn't changed that expected timeline. We just want to know the lineup and the price to expect. Considering a couple months savings could be the difference between a lower and higher priced product, announcement now would give time to price all components. But...

 

Intel Core i9-7900X Overclocked To 5.7GHz, Shatters Cinebench World Records



Intel Core i9-7900X breaks several benchmark world records


Intel Core i9-7900X Skylake-X Processor Tested Against Intel Core i7-6950X in CPU-Z Benchmark – Faster Than Intel’s Previous Flagship at a Much Lower Price


Intel’s Core i7-7800X 6 Core Achieves Cinebench R15 Performance Record – Fastest Hexa Core CPU To Date, 5755 MHz on LN2

 

Some nice overclocking there!

I think that it is good play on AMD - intel hits the market with the first batch with prices as per the slide, which suggests the pricing of i9s would be, again, as per the slide... then comes AMD with their pricing. Panic mode on (price reduction) again? We'll see.

 

See:
http://forum.notebookreview.com/threads/intel-i9-18-core-announced.805393/


Correction: there is zero information period from just; AMD.

Not arguing that i9 may have been announced a bit earlier than Intel was originally planning for...

The part I'm trying to point out is that it only seems 'sudden' to us - this was a long drawn out process and i9 would have been released eventually.

If AMD had an i9 in their back pocket, they would spring it on Intel in a second too - but that wouldn't make them any more magical than Intel either; they would have had to be developing it for a very long time already (and had simply not announced it to the world yet...) but they wouldn't be able to whip up something at this level in a few months - and there is no processor company that has 'almost/roughly finished' designs waiting on the backburner that they could trust to put in the limelight is such a short time either.

 

Exactly, AMD needs time to digest that newly announced Intel pricing and how it affects them.

If AMD had planned on releasing info, that Intel release delayed them - not necessarily due to any non-competitive reaction to their planned pricing - but there's just not enough "corporate" time to react.

I don't blame AMD for waiting, they are in the catbird seat watching Intel scramble, it's probably also good to keep Intel in the dark as long as possible.

We don't really care until it's close to time for AMD release anyway, Intel's gonna be a while delivering they new CPU's.

AMD's a small company doing a lot of things, so I don't expect a lot of notice when things release, especially with a big company like Intel ready to change their direction based on news from AMD.

 

AMD isn't keeping just Intel in the dark. It's keeping it's potential customers in the dark too - which is a far worse fate for a company.

See:
http://forum.notebookreview.com/threads/amds-ryzen-cpus-formerly-zen.799348/page-172#post-10537218


Did you miss that post above? Forget about the world records ('scores'=boring... yawn).

With an indicated ~22.5% increase in single core performance and a more modest ~3% multicore performance for an estimated ~$700 lower price point - Intel is delivering what customers what to hear: better performance than what they have now and lower cost too.

Both AMD and Intel are adjusting their previous outlooks because of what the other is doing.

Thinking that Intel is changing their direction on mere 'news' from AMD is wishful thinking at this point.

 

All this back and forth is making me want to hold off on upgrades entirely, I hope that's not a general market reaction and just me being picky and waiting to see what the aftermath looks like.

 

You have all reason to being picky!! It's your hard earned money. And you want most of every penny spent.

 

The First AMD X399 Threadripper Motherboards From ASRock!

Gigabyte X399 Aorus Gaming 7 Motherboard & VRM Specs

Gigabyte X399 Aorus Gaming 7 Motherboard & VRM Specs | Computex
By Steve Burke Published May 31, 2017 at 2:00 pm
http://www.gamersnexus.net/news-pc/2937-gigabyte-x399-aorus-gaming-7-motherboard-and-vrm-specs

 

Unless one has inside info - I doubt system builders are in the dark here it's the self builders that are in the dark.

*Yawns* and what does that equate to my bang for buck literally cpuing nothing. I fire it up and web browser or watch YouTube and poweroff end of the day. What multicore did that show off? Nada.... But new is the majority of users are you everyday of the mil users and all that performance is wasted when they need to do email, skype, www and word/excel that will never see the full potential of overpowered system that would cost then more and give them less in return. Only HardCore Gamers would push such system but the market is held up by the regular consumers that are more of the market then Gamers.

Why wouldn't one do that it's called competitions.

Actually they are - customer are not only CPU comparison but price as well and for the same cost a AMD can do what they need why spend more on more expensive iNtel that does the same thing. Your not giving consumer credits where it is due.

 

Well true, I guess I meant unnecessarily so. There's a point when it just becomes stalling and hurts more than helps.

 

AMD was teasing that they were going to give use more info and essentially left us high and dry with nothing additional. The blender demo was for naught without a comparison, more so than saying we did this before on a Ryzen chip.

And I agree, that totally felt like stalling. And it did hurt then in my eyes at least. From what I have read about the disappointment expressed by most AMD succeeded is upsetting allot of its fans. This with lack of Vega and ThreadRipper information.

And just a heads up, that Threadripper demo at 4K playing the game was 2x Vega's in crossfire was it not? I would have thought a single card could handle 4K gaming?

Edit; also what good is 4x graphic card slots where the ThreadRipper CPU's do not push the super fast cores needed to create the high frame rates having four cards could yield? So while there are a ton of PCI-e lanes it seems many will just sit there going to waste.

.

 

A lack of information isn't a call to jump to conclusions extrapolating data from zero data. It appears you we are letting our disappointment in AMD not providing solid information fill the gap left by the lack of information from AMD. It's way to earlier to give up on AMD.

There's nothing really to discuss except what AMD does announce and delivers.

The x399 seems like it's getting ready to deliver.

 

Not filling any information, I don't have any to fill anything with. Not extrapolating anything but I did find, not seeing the frame rates, it kind of suspicious running a 60Hz monitor at 4K smoothly but with 2x cards a bit awkward. I mean does not a 1080 TI do this with just one card usually?

Also I saw where they were pushing the 4 Vega's in the one demo. The fact allot of those introduced boards had 4 card capability. Ryzen because of single thread performance is not the best option for 2x SLI with 1080 TI's right now, do we think a threadripper will be better. Lastly unused PCI-e lanes just sitting there going to waste is just that, capability going to waste. Now maybe this is extrapolation but I think not far off the mark.

Edit; I am not giving up on AMD, they gave up on me! And yes I am upset about it if no one can tell.

 

That would seem to be AMD's point, differentiating Threadripper's capabilities being more expansive than Ryzen.

Pre-production hardware + drivers + game compatibility/crossfire support and all, it's not a fair comparison to the point of giving up - we've seen huge improvements in performance from games optimized for current AMD GPU's and Ryzen.

Also, those must be the slowest Vega GPU's, as the faster ones that compete with Volta aren't here yet.

Too many PCI-e lanes is a "problem" that Intel wishes *they* had

 

Sorry they opened the door for criticism, now they must face it. I'll apologize again as I do not feel sorry for them either.

And yes, Intel would love a 4 way SLI for 4K 144 Hz capability; http://www.144hzmonitors.com/monitors/asus-computex-2016-27-inch-4k-144hz-gaming-monitor/

 

The part I don't understand is why you think AMD promised everything you expected? What door? Did I miss something? Did AMD say they were going to do this, where did you read this?

Your expectations and (over) reaction seem out of place against what I've seen promised by AMD - not much.

And, we've had 4k monitors forever, a couple of generations of GPU's now, and they are just now being usable at gaming frame rates.

Just because vendors build 144hz/200hz 4k monitors doesn't mean we are going to get GPU's that can feed them concurrently with the monitor release - it could be years before we can drive those for AAA games, the same as when 4k @ 30hz/60hz monitors first came out.

 

The door is they were teasing us about new information concerning ThreadRipper and Vega, we got practically nothing that was not already known if you did some prior researching. What new information was there, nothing on these fronts. All that stream was about was just patting themselves on the back. They are for all intents just stalling.

TBH if they just ran multi-thread Cinebench on all the variants and reported the results I would have been very happy before, now I am not sure what they could do to bring me back, damage done. They in the end needed to get into some of the meat of the matter but did not even give us a crumb worth mentioning.

 

Two points: 1080 Ti only does 27fps min and 42-43 avg on ghost recon wildlands. Needing two for 60@4K isn't unheard of for even a Ti. So that may be an improper statement.

Second, TR cores were only at 40% during the 4 card demo on the max threads allowed by that program. (I think it was 28 threads shown in the example according to one article)

So, it isn't as simple as you suggest and no, the CPU was not a bottleneck there, most likely. Would IPC improved what they showed, unsure.

Sent from my SM-G900P using Tapatalk

 

That is the point, on the game demo, who knows what the real output was but 42-43 average is usually smooth and playable, so showing a demo with one card at 4K being smooth and playable would have been more impressive.

The second 4 card demo was a rendering program using the GPU's showing where the massive amount of PCI-e lanes have an advantage. that is not something we will get to see normally. While gaming may never use 4 way SLI or Crossover some rendering stations may be able too. I am just saying for most the massive amount of PCI-e lanes will never get used.

 

For 'most', you're probably right.

But this is one way to put PCIe lanes to good use:

See:
http://www.pcworld.com/article/3199...le-crazy-raid-configurations-for-a-price.html


With 8x 600p SSD's in RAID0 ~13GB/s (today) - in the near future ~128GB/s - with 20x Intel drives...

 

Now that I agree on. And one article pointed out tearing, which shows freesync wasn't used. Also, it could suggest the drivers need more work, etc. But, the largest critique is that, even if Nvidia needed two cards for the same performance, we have zero data to compare because they did not turn on the frame counter! That is just frustrating as Hell.

This is why I softened the stance on Vega. But I didn't on the CPU because of timing.

@Papusan - love the bench numbers. Noticed they are LN2, but were done at the event where LHe was available. I know the 7740K ones used helium. But, assuming LN2, the numbers look like BW-E numbers, not really able to bust through the 6GHz wall, but Damn impressive across the board.

As to the article showing the CPU-Z comparison between BW-E and SL-X, I saw no controls for the same speed on the single thread test. This means that the multithreaded test is more indicative of performance, which was low. By better utilizing boost 3.0 and likely using a higher speed on that thread, it shows nothing of the IPC performance, meaning I'd be a bit concerned, although it will kill it in games. Just worried on actual load.

Because the LN2 scores mostly were above the 1800X speeds, we can try to scale speed, assuming it is linear, then attempt to compare that scaled score, but lots of assumptions. Also, there is more development for the SL architecture than Ryzen, but that shouldn't be factored in.

So, even with the data we have, there are no clean scales to check out more info yet, even if derivative and not fully indicative.

Sent from my SM-G900P using Tapatalk

 

I never cared how it (threadripper) ran against Intel, LN2 or not. I wanted to see say scaling between core variants etc.. Even throw in a 1800x as base reference to show Quad memory channel etc. Run comparatives of the 1998x in quad and dual channel to show the advantage there. Give us some meat to chew on. There are plenty of one thing or another comparisons they could have put out to satisfy the inquisitive appetite. But what we got was a big fat slap on the back of the knuckles with a ruler.

 

We know it got 13.04 seconds in the blender test. Now, asking journalists to OC an 1800X to 4.0 and try with varying memory timings, then run BW-E and SL-X 8 cores to see scaling on IPC later this month, and finally the two Intel 8 cores overclocked on air or water would be nice.

Edit: So, to do an analysis of the 7900X versus the 8-core 1800x, I find myself unimpressed with the per clock IPC performance. Do not take this as gospel, as it applies in one case and use and is an estimate, not an absolute measure of performance.

Start by taking the Cinebench score and dividing the score by the MHz achieving that score. What you get is 3181/5755, giving 0.5527367506516073. This is then multiplied by the top score reached by the 1800X flagship, 5364MHz. These were benchmarks performed by the same overclocker, but it should be noted that the Ryzen was done at 3400 MHz as the AGESA update was not yet available, which means the BCLK had to be offset to achieve that score, while the Intel used 4000MHz ram, without needing to adjust the clock to achieve it. Also, now that faster ram is available with Ryzen, we may see faster IPC due to faster data transfers on infinity fabric. This IS NOT A PERFECT TRANSLATION OF SCORES. When you then multiply by the 1800X speed, you get 2964.879930495222. You then divide this number by the top score of the 1800X to get the performance over the 1800X. This gives 1.208182530764149, which means just under 21% increase in performance. But the analysis does not end there. We must look at the number of cores. When this is taken into account, then you see the 7900X has 25% more cores. This while achieving roughly 21% improvement at the same clock. When considering that the test on multithreaded performance of the SL-X on CPU-Z only beat the prior BW-E by 3%, we are starting to see that AMD is built better for specific multi-threaded tasks at an equal speed.

If instead, after reaching the score divided by speed, we divide the result by 10 cores, then multiply by 8 cores, then proceed by multiplying by the lower speed of Ryzen, we see a score of 2371.903944396177, which is 96.65460246113192% of Ryzen's score.

No 7820X score is currently available for comparison.

Of course, they don't reach equal speeds, as the Intel CPUs will achieve higher clocks than Ryzen. This has to be said, as you cannot just look at performance at a given speed and crown a king. Considering, on average, you can expect 300-500MHz more performance from the Intel overclocked, and this LN2 run is about that, let us run one more analysis only scaling to 8-Cores and seeing the ration between the two CPUs to look at a potential performance on similar core count. (3181/10)*8 =2544.8. To find out the performance over the 8-core Ryzen 1800X, you get 2544.8/2454=1.037000814995925. This means there is a 3-4% increase in performance, but that means we now consider pricing. The 7820X will cost $600 while the 1800X costs $500. This is 20% higher cost, while performance is 3-4%.

Now, some will point out that the percentage increase at over 5GHz is different than at 4GHz. That is correct. 500MHz over 4GHz is a 12.5% improvement in speed. 5.8GHz over 5.3GHz is only 9.43396%. This is a 3% difference in percentage of scaling. That is a fair criticism. So, you may see a slight widening in that score, while still being well under 5% on the above analysis including speed for average OC. This also must be noted that TR will have Quad Channel memory and this analysis compares a quad channel to a dual channel setup. Also, AMD now supports higher speeds and the scaling found on the data fabric. TR uses an even higher transfer rate data fabric, supposedly, with the cores using the Zeppelin rather than fully what was found on the Ryzen R7, if rumors are true. The Intel chipset is newer, but both tests were on unrefined systems, with the world record on the Ryzen 1800X performed at the February announcement for the press, not after release with all of the improvements that came within weeks of release. Further, Intel's architecture allows better utilization as Skylake architecture is not exactly new and software has already been optimized for it. Considering it is only 3% difference between the 1700 overclocked and the 1800X, while that costs almost 50% of what the Intel 8 core costs, that will continue to be my recommendation for the 8-core chips (unless release data shows this analysis to be considerably off base). If you need the quad-channel memory or high IPC, while only needing the larger core count for a couple tasks, there are situations where Intel will be the recommended go to processor, including the extra PCIe lanes for an 8-core, which will help drastically on storage support or network support. So please don't write Intel off on cost alone. If your needs are such that you primarily need IPC and PCIe, Intel is the way to go. But, depending on price and release of a 10-core chip by AMD, that may change as TR may be able to give you the extra storage you need. IPC still will favor Intel, but that will then be a smaller subset.

So, this is an analysis performed from public data from Intel. I will later try to find data on Intel's products running blender on the same image and compare the original Ryzen data to released Ryzen data running the blender image to Intel, while looking at the score on the TR, while noting we know nothing on ram speed or the speed that chip was running at in the Taiwan show. That is the most frustrating part of this. Hope this helps some with the comparison before numbers actually become available.

@TANWare @Papusan @Rage Set @triturbo @bloodhawk @hmscott @jaybee83

Sent from my SM-G900P using Tapatalk

 

We asked for it and we got it! Taken from the official AMD Ryzen facebook page! This is looking phenomenal! So, take this with my above analysis.

@TANWare

 

We can guess performance all we want, does not matter. They again could have within the family put out comparative benchmarks. That would be more useful. Prior to the disappointment all I cared about was how Threadripper would improve on Ryzen. Now that I have been turned off by AMD it does not matter at all.

Edit; as per the image on the facebook page that was 5/27/2017. So well before the stream we already knew this.

 

some new software update came out a day or 2 ago for memory kit compatibility. apparently XMP isn't as standardized as one would think so they added a tone of profiles or something.

 

AGESA 1.0.0.6. It is a microcode update and unlocked secondary and tertiary timings. This increased compatibility significantly. Still waiting on 4000 mhz screenshot.

Sent from my SM-G900P using Tapatalk

 

It wasn't official. That was the thing about Polaris, about Ryzen and now about Vega and ThreadRipper. Rumors are fun, but spoil the official information and/or release. It's up to the individual. I've learned my lesson with Polaris and dialed back for Ryzen and now pay attention to Vega and ThreadRipper. It's nice to have something to hang onto, but in a situation like this (obviously one-upping all the time), would you give away all of your cards? They have to get back some market share and this wont happen without some cat and mouse game. It's easier for me though, it's not like I have some $3-4-5000 burning my pocket to throw at desktop hardware, or need such computational power... for now. I heard that AM4 is still pretty good platform, so it would be one of these for me

For some odd (or not) reason I'm not getting notifications for a while now and I know that I'm not the only one.

Sigh. Would we EVER get better naming. I'm pointing at both directions here. Anyway, some nice base clocks, but I'm certain that they are once again clocked within an inch of stability.

 

The other reason on the clocks, this separates the dies. That may affect the heat spots on the IHS, which effects how hot it can be run, something which will likely effect Intel's speeds. I'm guessing OC will likely be similar to the R7/R5 chips as well.

But, even though my analyses aren't always correct, they tend to be very close to what the performance is. I'm hoping that holds, but will do a six core OC analysis/comparison for another data point, that way to compare and adjust this performance analysis, if needed (with official numbers for the 6-core after release later this month).

I don't know why it didn't give you guys notification, but was done as reading after the original post. Don't know if that matters....

Sent from my SM-G900P using Tapatalk

 

What about Ryzen mobile ?

I saw an Asus ROG powered by Ryzen 7 1700 (8 cores - 16 threads).


And a socket CPU :



I regret buying Intel's BGA now ...

 

That's gonna turn some heads.

 

3.0 Ghz huh, that's pretty good for an 8-core in a lightweight laptop. I wonder if it's overclock-able...

 

As long as ES can be bought in China ... not really a problem.

 

Good. Any idea on pricing of R9 1998? And its unlocked right?

3.8ghz is a very impressive clock speed.

 

Again,, they disappointed with the lack of information regarding Threadripper. Do I want them to give away the shop, no but better information that what we had would have been nice. The earlier board showings confirmed what was expected with the 64 PCI-e lanes and all those pins.

The only thing they really had was blender rendering but it did not show a comparison. and to be honest they did not tout it but where as the 18 core Intel may be the first with 1 teraflop on chip with 4x Vega FE's there is possible 100 Teraflops available on the x399 platform.

 

Really, really grasping at straws here, huh?

Okay:

• Choose a Number

• Add 5

• Double the result

• Subtract 4

• Divide the result by 2

• Subtract the number you started with

Bet you your number is 3, right?

There, I just proved AMD is best.


(Now you don't have to do the same silly analysis for Intel).