Even if Zen comes with SB performance, with DDR4 and the latest chipsets and controllers it will still be a very viable option. I'm sure many have upgraded from SB not because Skylake or Haswell offer that much more performance but because of all the goodies you get when you upgrade to these modern chipsets.
For AMD though it is a good thing still, they are moving to newer memory controllers which they have not done in years for example. Then they build off of Zen, kind of what Intel has done with Conroe and Nehalem. They're getting off AM3 finally. They're going DDR4. Sata controllers are going to be faster and efficient! Proper amount of PCIE lanes! Power draw should be lowered. APU's are going to get a lot faster! Unified socket design! No more FM2+, AM3+, and AM1! Zen might not be the fastest, but for AMD it's a start for them to improve. They need it too. Considering how much of a boost it'll be over Bulldozer, it's a big deal. Even more if the price point is perfect!
Very much so. I think it will *just* eclipse Ivy Bridge in overall performance, with a few areas where it really excels (tight integer-heavy loops, for example, which are currently the bane of the construction cores' existence). Traditional SSE and MMX performance should see a healthy increase - doubly so when talking about integer vector instructions. Floating point is very hard to predict, but if the alleged Cinebench results are accurate and actually represent a 100% boost from Orochi to Zen, I'd expect a good part of that to be from SMT and the removal of the CMT penalty present in Orochi. That would actually only equate to about a 50% FPU IPC improvement in that particular benchmark over Bulldozer - or about 35% over Excavator... pretty close to expectations. I'm assuming about 20% SMT scaling. Any less, and that IPC figure moves upward. 20% SMT scaling, for a first design, would merit AMD's design team a freaking award.
AMD has actually had DDR4 controllers for a long time - their cheapest CPUs have a DDR3L/4 dual controller, for example... AMD just doesn't have the platform on the desktop to allow us to use it. Just a minor pick ;-) As for SATA - I *really* hope they have good performance! I run 5x SSDs, so SATA3 performance is *very* important to me. I want my main rig to move to Zen + Polaris/Vega sometime next year (looks like it will be towards the end of 2017 now :bang:uke2. A 10% deficit for SATA performance won't mean much... unless it's in latency figures. But 20%... 30%... or what they have now... :micro::bolt::flame:
I do run 2 sata 3 SSDs, and 2 HDDs in Raid 0 + storage drive. I would not go for 3rd SATA SSD, I would get M.2/PCIe SSD instead to actually get something more than before. Drive performance is important because it is slowest part of PC caching system today. If you need more than you have... http://www.guru3d.com/articles_pages/toshiba_ocz_rd400_pcie_nvme_ssd_review,20.html That price for 512GB is not that bad considering performance.
I fully moved to M2 and PCIE drives not that long ago(Except for a bootloader on an SSD, and that is used for misc storage) and I notice somewhat of a difference when it comes to loading times. Performance in games and applications, there probably is some just not a big enough difference to notice.
It really depends on game/application. I have some which are big and are just fine on HDD, then there are some big/small which are far from OK on HDD, but shine on SSD.
Sandy Bridge performance. Doesn't double as a space heater. Priced between an i5 and an i7. If that's where Zen lands, it's good, but not the game-changer we were really hoping for. It would not alter the current pricing of CPUs much nor give i7 owners any compelling reason to consider AMD.
SandyBridge is the same as IvyBridge , are identical arhc.But SB was made on 32nm IB on 22nm . Moving from 32 to 22nm does mean at the same frequency the gain is 0% on speed , but you will have much less required power and yes Ivy has an iproved IGP over SB, that is the big difference between them. SB i7 2600k at 3.4Ghx and 95Watt and IB i7 3770k 3.5Ghz and 77Watt. Cinebench R15 - Multi Threaded i7 2600k 672 i7 3770k 708 So we have a gap here of 5% , if we set the 2600K to work at 3.5Ghz instead of 3.4Ghz we'll still have a gap of 2.3% and is not 0 , why? To answer at that question I had to read again the review for i7 3770k. The difference is that : the RAM of the system test for i7 3770k is Memory: 4 x 4GB G.Skill Ripjaws X DDR3-1600 9-9-9-20. Well SB had its memory controller limited to 1333 Mhz unlike IB that can run at 1600 Mhz . Cinebench R15 is not running inside the internal cache and is using a lot of system memory so yes IB could be 2.3% faster at the same clock than SB , not because are different just because are using different memory . Coming back to ZEN , it will bring a fresh new DDR4 memory controller , new instructions like FMAxxx, AVXxx atc, hypethreading etc ... and yes we are upgrading rigs for that kind of features , but sadly I still believe it will be in pure speed in the same class with Sandy Bridge (or Ivy ... is the same)
We are both right and wrong in the same time . I will give you an example . We have a SandyBridge with 6 execution ports and a Haswell that have 8 execution ports.We are just limiting at that example , nothing more. So SB 6 ports , 1/3 of that amount is 2 and that means 33% .Add 2 to 6 =8 you got a Haswell .... so Haswell has 33% more resources than Sandybridge. Now.... we have Haswell with 8 ports , 1/4 of them or 25% means 2 . 8-2=6 you got a SandyBridge Okay , but 33% is not 25% .... which one is the good answer? From a view point of a SB user, yes Haswell have 33% more resources,but from the point of view of a Haswell user , it have only 25% more than SB. Both of them are correct , but depends the system reference you take , that means relativity .
Why is it taking them so long to get this product into the market?! Maybe I am just use to seeing Intel deliver on a regular basis making Amd seem a tad/Alot slower lol. I have been reading about this Zen for over a year now and here we all are still waiting.
It's already sent to OEMs for verification. It will be out as a "back to school" product in the form of the first Bristol Ridge APUs (which I imagine will be something like the PS4 NEO in a smaller performance envelope), and then there will be the 8-core parts. It took Skylake a couple of months until CPUs were readily available and it wasn't a radical departure from the previous Intel designs, and Intel also has a ton of production capacity while AMD has none.
Ivy Bridge isn't exactly Sandy Bridge, but it is *very* close. It has some new instructions, updated microcode, updated memory controller, and a few very minor core changes (very very minor - caches, widths, units, stay the same). There is a performance benefit with Ivy over Sandy that goes beyond clockspeed. Some benchmarks show a rather consistent 10% or more improvement, but most are about zero to 5%. The reduced overclockbility, though, is a killer. I have no interesting in upgrading my 4.5GHz i7 2600k to an i7 3770k. Even if the CPU was free, I'd just sell it... after running it and doing a comparison, of course Sometimes you get lucky... but my i7 can run at 5Ghz (barely). Yes, Zen will almost certainly fall in that range of performance... but it should have areas of performance where it should trounce Sandy Bridge. I would not be surprised to see it match Sandy in FPU and stomp all over it in integer... just to lose again in memory (even with DDR4 3000 vs DDR3 2133). We will know... in time. :ying:
That's what I've heard, games that are built on UE I hear need an SSD to help out with texture streaming speeds. If I had the chance to do it again, I don't think I would. I would either just stay with an SSD/hybrid drive combo, or go RAID with SSD. I do like having less cables though, cleaner look!
Zen looks interesting, i hope for AMD's sake it will do well. I remember having the Athlon 800MHz Slot A cartridge and overclocked it to around 930MHz lol. A total beast at that time.