How would AMD impact intel if their Zen chips got the projected 40% better IPC over excavator and matched Haswell-E CPU performance?
I think this is going to make intel shit their pants and drive them to release cannonlake with more cores since they've lost the ability to significantly increment IPC.
I think at that amount over IPC it would be game over. Do you have a link?
The thing is, if they do THAT much of IPC increase, prices will go up significantly, and Intel, knowing their jew powers, will drop their prices to be a step ahead of AMD, simply because they can afford that
This would most likely cause AMD to become more popular and intel to lower prices in a desperate attempt to not go bankrupt.
It could also mean serious trouble for intel for the tablet market since AMD Zen SoCs might be more power efficient.
>>51970021
http://www.kitguru.net/components/cpu/anton-shilov/amd-zen-is-40-faster-than-current-x86-cores-zen-is-incoming/
Old but still relevant.
Intel isn't going to change anything they do. The X86 segment of the semi conductor market isn't about performance, power consumption, or any other hardware metric. Its about incentivizing OEMs to stock products with your hardware over the competition's, and intel already has the entire industry in a headlock.
AMD will regain enterprise market share which is a safe bet since they're currently only at 1%~ for servers at present. High end desktop chips aren't their bread and butter, and even strong sales here wouldn't change much unless their margins were huge. Though AMD themselves stated they did not see desktop PCs as a growth sector for them in 2016 so that pretty much tells you all you need to know. Summit Ridge CPUs will hit the market, they'll probably be priced against the i7E line on the high end, but their real mainstream parts won't launch till Q1 2017 with Raven Ridge. The APUs have the strongest sales potential with OEMs, and thats what AMD sees as the real cash cow for consumer grade chips.
How Summit Ridge actually performs is mostly irrelevant. Unless it totally outperformed Haswell-E/Broadwell-E its going to get nitpicked to death by mainstream reviewers. Headlines around the web will read "AMD scores big with Zen, but still isn't enough to dethrone Chipzilla" and that'll be the tone heard everywhere aside from AMD fans.
Enterprise, mainstream APUs, and semi-custom designs are the foundation for post restructure AMD. Even if they're competitive against high performance intel parts they'll still have tons of spin thrown against them for not winning out in every serial bench.
>>51970202
Their chips would be cheaper but not shit. This would peak the interest of OEMs as they look for more ways to make more shekels.
An 8 core Haswell IPC CPU would be very competitive if priced right.
If it's priced at FX price levels it would be a no brainer.
>>51970183
Thanks.
>>51970202
So they are fighting on Intel's turf more or less? Would you say custom built sales are comparable to OEM sales? I don't have a clue what I would search for to find that info
Zen is AMD's swan song.
AMD revenues have been down these past two quarters. AMD is kill in 2017.
dont believe their lies
an overclockers dream
>>51970046
yeah, amd will need at least 3 or 4 gens of consecutive superior cpus to balance the game.
The only way to beat intel dumping is by a long term attrition.
>>51970304
As far as semi-custom sales go, they're not bad. They've sold well over 40,000,000 chips for the PS4 and Xbox One since they launched. For some perspective on those numbers In 2012 and 2013 AMD was selling around 60-70 million CPU/APUs total each year. Semi-custom design wins have huge potential for making their bottom line, and they're coming out with a few more next year.
More so than fighting on intel's turf they're carving out niches for themselves. Intel is a company who can generate $50,000,000,000 a year in revenue, and AMD is a company making about 1/10 that. They just don't have the finances, supply chain, or clout to directly compete against intel in major market segments.
>>51970002
'no'
>>51970002
It wouldn't do jackshit. Even with a 40% increase they would still be way behind in IPC. And intel would simply undercut the fuck out of their prices just like nvidia does whenever amd poses a slight threat because they can afford to. AMD has no chance, but at least the consumer wins.
>>51970626
this
broadwell is about 5% overall faster than haswell. skylake is overall 10% faster than haswell.
haswell-e is haswell without the integrated gpu and extra core count.
if zen matches haswell, amd will still be two generations behind in ipc performance.
granted, a MUCH better state than they're in now. but still slower nevertheless.
and by the time zen lands, intel will be releasing skylake's refresh, kaby lake. granted it won't produce a ipc increase, but gives intel a chance to update the chipset (yeah, i know), bump up stock clock speed, and even core count if they wish.
not only that, but guess whats coming out next year as well? broadwell-e. broadwell-e will give the same core count amd will be giving zen, but with faster ipc.
>>51970002
It wouldn't matter if it was 40% higher than Skylake. Don't you remember what happened back when AMD was beating the shit out of Intel?
Intel still won through marketing and underhanded tactics.
>>51970733
If they released something with Haswell performance and 8 cores + SMT for a few hundred Dollars, they've definitely got a great deal.
Also, no way in hell Broadwell and Skylake are 5 and 10% faster than Haswell. It's less than that.
>>51970002
>matched Haswell-E CPU performance
No way in hell are they catching up that easily. If they do, however, then I'm going to get an 8 core with SMT and call it a year. They better hurry the fuck up, I put the 5820k on hold when I heard about Broadwell-E coming out with an 8 core variant that isn't the highest tier.
>>51970957
This. I believe Skylake is a nice 20% faster than Sandy bridge, but that took several years so it's pathetic. If I remember correctly, Skylake is a mere 3% faster than Broadwell.
>>51970957
Broadwell was a 1-3% uplift over Haswell, and Skylake is like 2-5% faster than Broadwell while also magically managing performance regressions in a few areas.
>>51970957
http://arstechnica.com/gadgets/2015/08/intel-skylake-core-i7-6700k-reviewed/
>Therefore, it's disappointing to see slim performance gains of between 8 to 16 percent from our testing.
>8 to 16 percent
so yeah, it is.
>>51970957
it really depends, but overall, 10% is a general average, though some go lower and say 8%. some areas its as little as 4%, while others its 20%. like video encoding for example where the gains are some of the highest since sandy's release over nehalem.
zen won't cause intel users to switch in droves. it will just cause amd users to remain amd users.
some nehalem might switch for wanting a change in face, possibly some sandy users for the same reason, but overall, zen isn't that attractive to intel users since.... intel already offers and surpassed (as small as it may be) zen performance since haswell's release in 2013.
40 percent over their current parts, not Intel's.
40 percent isn't that big a deal considering that their IPC is miserable, and their clocks are extremely aggressive, meaning the thermals are dreadful. This only approaches Intel's IPC several years after the fact, and on top of that Zen is not landing until the end of 2016.
>>51970914
AMD's own lack of competence also mean that they suffered much more that just what Intel did to them. It all went to shit for them from 2006.
>>51970157
>This would most likely cause AMD to become more popular and intel to lower prices in a desperate attempt to not go bankrupt.
But that's fucking wrong. Do you understand how much money Intel makes beyond the consumer market arguing about muh gaming cpu?
>It could also mean serious trouble for intel for the tablet market since AMD Zen SoCs might be more power efficient.
Kek'd
>>51970387
Then Intel can box up whatever dogshit CPU they want and charge anything for it, and to continue running x86, we'll have to buy it. Fucking joy!
>>51970387
AMD won't go belly up just yet. Hopefully Zen will be really, really awesome, enough that they can get past the many hundreds of millions in bonds.
>>51971964
You mean like they are doing now?
>AMD's next chip going to be 209823095235x betterer than Intel's latest.
>What is every AMD's promises every release ever?
It's hilarious how you neckbeards think every fucking AMD release is the second coming of Jesus, and every time, it turns out to be shit.
>>51972322
>i like to generalise to feel elite
>>51972322
The 290x kicked major ass when it first came out senpai
>>51972457
This. Is was extremely expensive when it came out, but it was cheaper and a better performer than the 780. The 780Ti did change things a bit, even though that's not the case right now.
>>51972457
4870 was pretty sweet when it came out, undercut the gtx 280 hard. Then the mining craze hit, just like LTC and the GCN cards.
I really do hope its that good just for the sake of competition
>>51970626
Yet, this hasn't happened from Nvidia despite the insane bang to buck value AMD is providing with current line up of graphics card.
