AMD Zen Performance Preview

drescherjm said:
I fully expect Zen will be a success if the 40% IPC (putting the IPC somewhere between Ivy Bridge and Haswell) that AMD has stated is accurate.
Click to expand...
as much as I'm rooting for AMD I have a sinking feeling that that 40% is going to turn into 40%*(*up to 40%). but even if they can get close and come in lower priced, zen should still sell decently.
 
ChadD said:
AMD shouldn't have to many issues with cash... they aren't going away anytime soon. When it comes down to it plenty of large investors are still willing to sink cash into AMD. Cause lets face it if they ever did go tits up their IP alone would be worth a fortune. I'm not saying the have the cash on hand to go to war like Intel does though. Should be a fun Jan-Feb 2017. If zen turns out to be what the leaks are claiming and performs = to Intel chips selling for 2x the price. Even if it does't sell like mad it will keep investors happy. Also I know I mentioned AMDs ARM ambitions, I think that still has plenty of investors happy on that front as well. Lots of people still betting on ARM servers to take off at some point and AMDs optron name still has some weight in that market. Even Polaris... as an investor I may not really care that Nvidia is selling more gaming cards that have razor thin margins. I'm looking at that slimmer gpu design and thinking AMD is considering it for SOC chips. At this point AMD could enter the high end phone market by simply retooling the Optron A1100 core and slapping on a Polaris GPU. The Polaris based cards may be low ball for PC VR... but a AMD ARM chip with a Polaris GPU in an android phone? Why not they have the tech and Samsung is clearly a friend in regards to the fabs. Both companies are going to be under some pressure this year... who knows I wouldn't be shocked to see them announce a High end Samsung phone powered by and AMD-ARMtron / Polaris Samsung Fabbed SOC.
Click to expand...


Its not purely a cash thing, yeah cash will keep the afloat, but how do you take the fight to Intel, can't do it with monetary push, just not possible, people are interested in AMD stock prices, those investors don't give a crap about sinking money into AMD because AMD needs to be solvent, which right now they are not, and if Intel responds to Zen the way I expect them too, they will not stay solvent. Damn even with much better products AMD couldn't hurt Intel. Intel's capabilities are much more then what they are making on the books, their infrastructure is what is most important to them. If you look at what Intel has, it like MS just bigger because the multiple billions they have in infrastructure. No amount of AMD IP is worth that much. This is why you don't poke a sleeping Gaint, cause many things can happen that you can't foresee. AMD isn't in control of its own future, Intel has complete control even if they give AMD 3 years of respite.

ARM doesn't make sense at least on when they have a good x86 architecture, this is why Zen Arm has not been talked about recently, something changed in AMD's view for ARM products, what that is don't know.

You are looking at AMD with what 3 billion worth vs a company Intel which is worth 150 billion? There is a monumental difference is Intel's capabilities.

This is why AMD can't go head to head with both Intel and nV, its just not possible even if products are similar in performance. They need to have better products then their competition or find areas where they don't directly compete with those companies.
 
Last edited:
  • Like
Reactions: ChadD
like this
razor1 said:
Its not purely a cash thing, yeah cash will keep the afloat, but how do you take the fight to Intel, can't do it with monetary push, just not possible, people are interested in AMD stock prices, those investors don't give a crap about sinking money into AMD because AMD needs to be solvent, which right now they are not, and if Intel responds to Zen the way I expect them too, they will not stay solvent. Damn even with much better products AMD couldn't hurt Intel. Intel's capabilities are much more then what they are making on the books, their infrastructure is what is most important to them. If you look at what Intel has, it like MS just bigger because the multiple billions they have in infrastructure. No amount of AMD IP is worth that much. This is why you don't poke a sleeping Gaint, cause many things can happen that you can't foresee. AMD isn't in control of its own future, Intel has complete control even if they give AMD 3 years of respite.

ARM doesn't make sense at least on when they have a good x86 architecture, this is why Zen Arm has not been talked about recently, something changed in AMD's view for ARM products, what that is don't know.

You are looking at AMD with what 3 billion worth vs a company Intel which is worth 150 billion? There is a monumental difference is Intel's capabilities.

This is why AMD can't go head to head with both Intel and nV, its just not possible even if products are similar in performance. They need to have better products then their competition or find areas where they don't directly compete with those companies.
Click to expand...

You are right no doubt. Of course Intel acting the big boy and squashing AMD... still requires them to act smart. I'm not sure Intel is going to do much about Zen. I don't see them dropping the prices on their flag ship parts 50% or even 25%. At least not day one or anything. Perhaps if Zen does start selling the do that... we'll have to wait and see if Intel is going to play hard or not. On the arm stuff your right AMD talked alot about it and has been pretty quite lately... the A1100s are still shipping though so they are hardly out of that game. I believe they are just biding their time in that.... put a toe into the lower power server market. It gives them a completed high end design that they can build off for other stuff later. I mean intel is doing the exact same thing now... they announced they are creating a Intel Arm chip for phones to ship in the next year or so. I think AMD choose the right way around, it wouldn't be hard to strip a few of the higher end features out of their Opteron design stuff like memory ECC ect and have a very nice performing phone chip if the market is there. I do believe in a few years though ARM will be headed to the big time in main stream consumer desktops. Anyway my only point there was saying by having their toes in a market like that it keeps a lot of investors happy... people interested in banking on low power servers and those thinking way down the line to possible higher end ARM parts for new devices, AMD has done what they need to at least keep the investors on their side for now at least.

People counted Apple out at one time... I remember after 7-8 years of disaster and ever declining sales. It took one product to turn it around for them fast. AMD won't make their own turn around device. Your right Zen is a desktop processor its not a disruptive product like a Ipod. Still, AMD is in a good place to power a product that could make a difference. I'll go even more crystal ball and pull something out my backside. Samsung could use a hit... and I think its a safe bet to assume everyone is working on Augmented reality devices. I know they make the Exynos chips. Still they could generate a lot of hype throwing a AMD CPU/GPU into a phone or even small desktop with AR built in. A Samsung version of a hololens type laptop/mini desktop. Anyway know I'm talking about them needing another down on their luck company to save them... damn sell the stock. ;)
 
Intel has been already dropping prices more than 50% on Xeons with consumers and volume discounts, Zen isn't going to change that, the only thing Zen will change is the MSRP prices we see on ARK, which well quite frankly don't reflect what Intel is selling them at. Now I do expect them to drop their 6 core parts down to what Zen's 8 core parts are at, but that's not much of a drop looking at 50 bucks (10%). Another drop for motherboard prices for those chips though. We know Intel plans for 6 core Kaby Lake to fit into the same socket as the other 4 core ones, but this will only by for Purely I think or what ever they call the desktop variant.

Apple is going back down now again, the reason why Apple came back is because they innovated and created a product that had no competition, the iphone. Added to that with their itunes cornered the market for music devices. Without innovation they die, pretty much without Steve Jobs. Look at what is happening right now, just last week Apple cut down orders for iphone 7, they aren't selling as well as they thought they would. In a few more years if Apple doesn't' make a new product that is innovative, you will see them back to where they were. And its not going to be pretty for them, MS is carving away in ipad market (which they over took apple already) and PC space apple has, other companies are slowing chipping away with their "premium" product lines, and MS is also leading into that with their surface line.

If AMD can't fix their power usage on their GPU, they won't be using that tech for mobile lol. Does AMD really want to go against Qualcomm, Samsung? Can they those companies shut out nV even though nV's tegra was better in graphics.....
 
razor1 said:
If AMD can't fix their power usage on their GPU, they won't be using that tech for mobile lol. Does AMD really want to go against Qualcomm, Samsung? Can they those companies shut out nV even though nV's tegra was better in graphics.....
Click to expand...

Again don't disagree with you Apple is cooked if they don't do something new. Just saying 3 billion 30 billion doesn't matter in consumer electronics. Comebacks happen... staying on top isn't easy for anyone including intel. They stopped worrying about AMD sure, I'm not sure its AMD they really need to worry about anymore of course. Just had to post though, cause Tegra had other issues and they where processor related. The GPU was fine the rest of the chip jobbed power. Of course NV isn't done with that chip... looks like they may sell plenty with Nintendo (of course AMD sells with the other guys for whatever its worth). NV may have found a niche for their chips with the automotive stuff. Anyway just threw it out there. I do believe AMDs arm stuff was just a toe in the water. Looked good for investors, and its a good way to hedge some bets. If Arm does start moving up the device chain past chorme books. AMD is better off having already built a chip then not. Or of course low power servers could start selling in larger numbers. Goal number one for AMD is zen though your right... Zen or bust pretty much at this point. If it flops it will be interesting to see where they go, could be fun really some desperation parts might create something truly ground breaking. (that's where I was going) Samsung is under pressure to be broken up and they have had a bad run of press on some products. If AMD strikes out, the two could make some strange desperate partners.
 
comebacks of Apple's magnitude are rare, Apple *Jobs in particular* had a vision, a vision that no other company could see at that particular time and with the itunes ecosystem which pretty much became inclusive of apps for the iphone too and then creating a cell phone from what was to be an ipad ( the reason why it wasn't feasible at the time to create an ipad, it was cost prohibitive) but as a phone the tech worked cause they could charge much more through monthly plans, hard to find things like that man. No other company was in that market, so Apple just took everything from them, the product was so much better then any other phone at the time, it can't be stopped. Just like Athlon, 64 and x2's, they had the same advantage being in a league of their own.

nV didn't find the niche for AI, they made that niche they too had vision, they saw it coming from years before, but it took since g80 to now to make the market see it, that is what costs resources to do something like. Opening up a new market costs money. nV did it slowly little by little. AMD doesn't have money to do this, and when they did, when they were ATi and AMD were separate, both companies squandered their lead.

The problem with ARM is its a completely new market for AMD, with high stake players in it, and that is something AMD needs to figure out, how to go into it without disrupting the current companies too much so they don't get their attention. nV went in bull headed, and it didn't take long for them to get pushed out, even though they are about quarter the size of qualcomm, they get crushed, AMD is much smaller than that 1/8 size.

Qualcomm also tried to create high performance ARM CPU's to go against Intel, that didn't work so well at least not yet. And this will not change for quite some time till they have the software to back up their chips.
 
razor1 said:
This is why AMD can't go head to head with both Intel and nV, its just not possible even if products are similar in performance. They need to have better products then their competition or find areas where they don't directly compete with those companies.
Click to expand...

Agree. AMD has done well in semi-custom, this is a market that Intel and NVIDIA largely ignore and it is one in which AMD now has a good track record. It is also a profitable business for AMD, but unfortunately the respectable operating profits that this biz generates are somewhat wiped away by AMD's operating losses in Computing and Graphics.

AMD has even been cutting back its spending in Computing and Graphics, but that business is in such feeble shape that it's still a money loser.
 
Not sure what anyone else thinks- but I'd suspect a major leak of benchmarks in the next 30 days. This is something to look forward to no matter how you feel about AMD's past performance.

This is an exciting time. Let's enjoy it. When I think of other product launches that I've either participated in or watched as a spectator it's a welcome feeling of Deja Vu combine with excitement.

So we await the benchmarks. And we'll react accordingly.
 
I am very much looking forward to this. Also, has anyone else noticed that when you run a single core test with Cinebench, it does not max out a single core at all? Basically, I noticed that it spread the work load across all 8 cores of my FX processor which basically means it is not a single core test at all. (Just a low load multicore test.)
 
TaintedSquirrel said:
Okay so what's the difference between Ryzen, Zen, and Summit Bridge? Too many codenames.
Click to expand...

Well, the final products havent been publicly announced yet, so we should kind of expect codenames and for stuff the be cleared up more once they do.

I suspect we'll know a lot more tomorrow :p
 
ManofGod said:
I am very much looking forward to this. Also, has anyone else noticed that when you run a single core test with Cinebench, it does not max out a single core at all? Basically, I noticed that it spread the work load across all 8 cores of my FX processor which basically means it is not a single core test at all. (Just a low load multicore test.)
Click to expand...

Nope. That's not how it works.

It actually IS a single core test, but the scheduler moves it around from core to core to spread the heat evenly so fast that if you look at loads it looks like each core gets an average low load at about 12.5%

But this average load is not like typical loads. It's actually thatthe CPU is pinned at 100% 12.5% of the time, and sits idle the rest of the time, and it averages out to 12.5%
 
Presbytier said:
That is always a possibility, but if the design holds up it is a pretty interesting chip and there is no reason to assume those are fake slides on the eve of the official reveal.
Click to expand...

Oh I meant Choo Choo train as in "Everyone aboard the Hype Train!" . I don't see anything wrong with the slides; they look quite interesting. I'm brainstorming how to fool the OC tool with chilled water. ;)
 
cageymaru said:
Oh I meant Choo Choo train as in "Everyone aboard the Hype Train!" . I don't see anything wrong with the slides; they look quite interesting. I'm brainstorming how to fool the OC tool with chilled water. ;)
Click to expand...
I imagine that has to be temp regulated and mobo voltage regulated.
 
XFX alluded to the RX 480 OC's depending on how stable the voltage is on the card. Of course they then went into a speech about how they used better than factory electronics / cooling on their card. I wonder if motherboard manufacturer will matter for Zen as in the quality of the components used?
 
cageymaru said:
XFX alluded to the RX 480 OC's depending on how stable the voltage is on the card. Of course they then went into a speech about how they used better than factory electronics / cooling on their card. I wonder if motherboard manufacturer will matter for Zen as in the quality of the components used?
Click to expand...
I dare say it probably does, being able to properly judge the voltages the mobo is capable to output has to be a must here.
 
SighTurtle said:
http://videocardz.com/64741/ryzen
Click to expand...

In that slide deck I certainly hope "3.4ghz+" means that the lowest SKU starts at 3.4Ghz and there is a 4.2Ghz top model, otherwise this is going to be a Bulldozer-sized disappointment all over again...

There is no way that 3.4Ghz plus their adaptive overclocking turbo thingamabob is going to be fast enough to be interesting to us, even if they to live up to their Excavator +40% IPC estimate.
 
Zarathustra[H] said:
In that slide deck I certainly hope "3.4ghz+" means that the lowest SKU starts at 3.4Ghz and there is a 4.2Ghz top model, otherwise this is going to be a Bulldozer-sized disappointment all over again...

There is no way that 3.4Ghz plus their adaptive overclocking turbo thingamabob is going to be fast enough to be interesting to us, even if they to live up to their Excavator +40% IPC estimate.
Click to expand...

Why? I am tired of hearing 3.4ghz stock is going to be a disappointment. Serisously its an 8 core processor if it is indeed 40% higher IPC then it is pretty good. Look at intel, they don't push their processors with higher core counts alot higher at stock either. if it overclocks to anything 4ghz, I say thats a win on first go at a brand new architecture.
 
NKD said:
Why? I am tired of hearing 3.4ghz stock is going to be a disappointment. Serisously its an 8 core processor if it is indeed 40% higher IPC then it is pretty good. Look at intel, they don't push their processors with higher core counts alot higher at stock either. if it overclocks to anything 4ghz, I say thats a win on first go at a brand new architecture.
Click to expand...

Well, I have a geriatric 5 year old hexacore Sandy-E 3930k at 4.8Ghz.

If I can't get at least par performance in single threaded benchmarks as this 5 year old chip, I simply can not buy one.

I have no need for 8 cores. Heck I barely even use the 6 cores I have. What is paramount is the highest possible single threaded performance.

AMD should have learned their lesson by now about trying to make up for slow cores by just dumping on more of them.

I am not expecting them to catch up in per core performance with Kaby Lake or Skylake or even Haswell, but if they can't surpass the per core performance of my 5 year old Sandy Bridge-E then Zen is an absolute turd. I mean, my SB-E chip is 32nm for crying out loud. Zen's process is freaking less than half the size.

If it can't even do that, you could give it a million cores, and it still wouldn't make up for it. If that winds up beibg the case, I'd call it a bigger failure than even Bulldozer.

Bulldozer, as bad as it was, did not trail 5 year old chips in per core performance when it launched.
 
Last edited:
Zarathustra[H] said:
In that slide deck I certainly hope "3.4ghz+" means that the lowest SKU starts at 3.4Ghz and there is a 4.2Ghz top model, otherwise this is going to be a Bulldozer-sized disappointment all over again...

There is no way that 3.4Ghz plus their adaptive overclocking turbo thingamabob is going to be fast enough to be interesting to us, even if they to live up to their Excavator +40% IPC estimate.
Click to expand...

Top sku on "Rice" seems to be 3.2Ghz base, 3.5Ghz turbo.
 
Zarathustra[H] said:
Nope. That's not how it works.

It actually IS a single core test, but the scheduler moves it around from core to core to spread the heat evenly so fast that if you look at loads it looks like each core gets an average low load at about 12.5%

But this average load is not like typical loads. It's actually thatthe CPU is pinned at 100% 12.5% of the time, and sits idle the rest of the time, and it averages out to 12.5%
Click to expand...

Thread allocation is the domain on the OS, not the hardware.

On Windows, thread allocation is simple: For all runable threads, the ones with the highest current priority ALWAYS run. So what happens is some higher priority thread eventually bumps Cinebench's thread from whatever core it is running on. When it gets rescheduled, it gets put on whatever CPU core is either free or running the lowest priority thread, which may not be the same core it was last running on. As a result, on Windows, threads tend to bump between CPU cores unless specifically forced to a single core by the application/user intervention. That's why you don't see a single core go to 100%, as the thread(s) Cinebench is running aren't locked to the same CPU core over a one second timespan.

On Linux, runqueues are used instead, which tends to lock threads to specific cores unless the load-balance algorithm decides threads need to be shuffled around to better balance the load. The downside to this approach is overall system throughput suffers, as there will be periods where runnable threads will not be scheduled by the OS.
 
gamerk2 said:
Thread allocation is the domain on the OS, not the hardware.

On Windows, thread allocation is simple: For all runable threads, the ones with the highest current priority ALWAYS run. So what happens is some higher priority thread eventually bumps Cinebench's thread from whatever core it is running on. When it gets rescheduled, it gets put on whatever CPU core is either free or running the lowest priority thread, which may not be the same core it was last running on. As a result, on Windows, threads tend to bump between CPU cores unless specifically forced to a single core by the application/user intervention. That's why you don't see a single core go to 100%, as the thread(s) Cinebench is running aren't locked to the same CPU core over a one second timespan.

On Linux, runqueues are used instead, which tends to lock threads to specific cores unless the load-balance algorithm decides threads need to be shuffled around to better balance the load. The downside to this approach is overall system throughput suffers, as there will be periods where runnable threads will not be scheduled by the OS.
Click to expand...

Yep, I didn't know the finer mechanics of how and why the scheduler does what it does, but the outcome is the same.
 
Shintai said:
Top sku on "Rice" seems to be 3.2Ghz base, 3.5Ghz turbo.
Click to expand...

This is a damned shame, unless it overclocks well, and by well, it needs to git just north of 5ghz.

We know to expect lower IPC than even previous gen Intel chips (simply by looking at AMD's Excavator + 40% claims) which sound out it at a Sandy to Ivy level somewhere, which means it needs to make up for this deficit in clock speed.

But what about Broadwell, some say? Broadwell is also a 8 core chip and it has lower clock speeds.

Truth is AMD doesn't have that luxury. They don't have a 4 core Skylake/Kabylake for more mainstream desktop/gaming applications where per core performance is key.

Zen needs to fill both these roles and it's going to fail abysmally if the per core performance can't be broalught up.

The truth is, except for a few specialty use cases (encoding/rendering, and I don't know about you, but I can't even remember the last time I did this) on desktops/workstations more than 4 cores is a complete waste. It's ALL about the per core performance.

The only saving grace when it comes to Broadwell-e is its overclockability. At stick clocks, I would never buy a broadwell-E chip. (It's just a shame you get so few PCI-E lanes on the non-E parts)

If Zen can't overclock to 5+GHz it will fall behind my geriatric 32nm hexacore Sandy-E at 4.8Ghz and will be a complete and total fail in my book.
 
I guess we will see where the IPC puts them. I hope this is a good chip from them and I hope stuff improves. Could always look at the lower clock speeds as the athlon 64 vs p4 days :)
 
Bah! Till Kyle's review comes out, we should all just stay at the station and quit jumping on the train.
 
luke51087 said:
I guess we will see where the IPC puts them. I hope this is a good chip from them and I hope stuff improves. Could always look at the lower clock speeds as the athlon 64 vs p4 days :)
Click to expand...

True, and I agree, until we have reliable 3rd party reviews from trusted sources like Kyle and possibly Anandtech, we really know little more than nothing, especially considering how AMD contorts themselves to cherry pick benchmarks and use their own very specific controlled settings to put themselves in as favorable a light as possible.

That being said, the only official thing we have from AMD thus far is that Zen has an IPC of Excavator + 40%. That puts them behind Ivy Bridge, so they have to make up for it in clocks.

Could they beat their Excavator +40% goal? I guess, but knowing AMD they generally fall behind estimates more than they beat them, at least in the last decade.
 
Zarathustra[H] said:
This is a damned shame, unless it overclocks well, and by well, it needs to git just north of 5ghz.

We know to expect lower IPC than even previous gen Intel chips (simply by looking at AMD's Excavator + 40% claims) which sound out it at a Sandy to Ivy level somewhere, which means it needs to make up for this deficit in clock speed.

But what about Broadwell, some say? Broadwell is also a 8 core chip and it has lower clock speeds.

Truth is AMD doesn't have that luxury. They don't have a 4 core Skylake/Kabylake for more mainstream desktop/gaming applications where per core performance is key.

Zen needs to fill both these roles and it's going to fail abysmally if the per core performance can't be broalught up.

The truth is, except for a few specialty use cases (encoding/rendering, and I don't know about you, but I can't even remember the last time I did this) on desktops/workstations more than 4 cores is a complete waste. It's ALL about the per core performance.

The only saving grace when it comes to Broadwell-e is its overclockability. At stick clocks, I would never buy a broadwell-E chip. (It's just a shame you get so few PCI-E lanes on the non-E parts)

If Zen can't overclock to 5+GHz it will fall behind my geriatric 32nm hexacore Sandy-E at 4.8Ghz and will be a complete and total fail in my book.
Click to expand...

40% IPC over Excavator would put it in the range of Skylake.
 
40% IPC over Excavator would put it in the range of Skylake.
Click to expand...

I would expect closer to Haswell than Skylake. Although if the gain is truely 40% AMD may be only 10% to 20% behind the latest from Intel on IPC. However they also are expected be behind in frequency (at least with the 8C / 16C compared to Intel's 4C / 8T).
 
variant said:
40% IPC over Excavator would put it in the range of Skylake.
Click to expand...

Not even remotely close. Skylake on an IPC basis is around 100% over Pilediver.

I wouldn't be surprised if Zen ends up as a 8 core SB.
 
variant said:
40% IPC over Excavator would put it in the range of Skylake.
Click to expand...

drescherjm said:
I would expect closer to Haswell than Skylake. Although if the gain is truely 40% AMD will be only 10% to 20% behind the latest from Intel on IPC. However they also are expected be behind in frequency.
Click to expand...

Both of you overestimate the IPC of Excavator.

I've done the math based on Anandtech's single threaded scores (because multithreaded scores just convolute things) in Cinebench R11.5.

They don't have any Excavator scores, so I took Kaveri scores, divided by max turbo clock, multiplied by 1.05 to make them ~Excavator, and then multiplied by 1.4 to approximate Zen.

The results wound up falling ahead of Sandy Bridge, but behind Ivy Bridge (when similarly taking their scores and dividing by max turbo clock.

I used max turbo clock rather than base clock because it's a single threaded benchmark, so the chips should be hitting their max turbo clocks.

I used Cinebench 11.5 only because it is the only single threaded benchmark I have been able to find scores for a wide variety of CPU's in, courtesy of Anandtech.com/bench...

Using multithreaded benchmarks makes no sense as between differing levels of scaling with multiple cores, as well as logical cores vs physical cores and the like, as well as potential memory bandwidth issues, the IPC winds up being obfuscated.

What methodology have you guys used when doing the IPC calculations?

My 5 year old SandyBridge-E overclocked on water at 4.8Ghz scores 1.92 in Cinebench 11.5 single threaded. I'll be looking for Zen to meet or beat this score (at max overclock on water) if I am going to buy it.

I'd be willing to do a performance side-grade in order to get more modern features and a motherboard that's not falling apart, but I can not accept a performance downgrade compared to my 5 year old ancient system.

I don't give a rat's ass about the extra two cores. In real-life use, once you get to 4 cores or above, higher single thread performance beats having more cores 99.99% of the time.

Unfortunately, according to my calculations above, the overclock would have to be pretty extreme on Zen in order to reach that score, so I am kind of doubtful it will happen :(

I really want to buy AMD again, for old times sake, and to support a healthy competitive market, but I can't justify this if a brand spanking new 14nm part can beat an ancient 32nm part.
 
Last edited:
I'm so freaking impressed by the numbers that i am seeing on these previews. Quite honestly i don't think that anyone, not even on their wildest dreams, thought that AMD had a real chance to getting close to Intel's 6k series.
 
Revdarian said:
I'm so freaking impressed by the numbers that i am seeing on these previews. Quite honestly i don't think that anyone, not even on their wildest dreams, thought that AMD had a real chance to getting close to Intel's 6k series.
Click to expand...

Well, keep in mind it's AMD's marketing department that is bringing this to you. They have a long history of extreme cherry picking of benchmarks pre-release under unusually controlled circumstances, making their products look WAY better than they wind up being when reviewers get their hands on them.

This may be good news, but it also might mean absolutely nothing at all. We will see when real reviewers get their hands on them.
 
i don't think that anyone, not even on their wildest dreams, thought that AMD had a real chance to getting close to Intel's 6k series.
Click to expand...

I thought the 8C / 16T would be close to Intel's 6C / 12T in heavily threaded code and would be priced as a discount from the lowest end Haswell-E.
 
Revdarian said:
I'm so freaking impressed by the numbers that i am seeing on these previews. Quite honestly i don't think that anyone, not even on their wildest dreams, thought that AMD had a real chance to getting close to Intel's 6k series.
Click to expand...

You can download and run the test yourself.
http://www.amd.com/en-us/innovations/new-horizon

My 6700K uses ~65W running the bench and not ~90W. ;)
 
drescherjm said:
I thought the 8C / 16T would be close to Intel's 6C / 12T in heavily threaded code and would be priced as a discount from the lowest end Haswell-E.
Click to expand...

Yeah, that was more or less my anticipation as well. Slightly worse lightly threaded performance, slightly better multithreaded performance than the 6800k. If it can even stay in the ballpark of the 6900k in the real world, that's an impressive achievement.
 
You must log in or register to reply here.
Back
Top