• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

*** AMD "Zen" thread (inc AM4/APU discussion) ***

Intel's 8+ core CPUs are massively overpriced due to no competition and an effective monopoly. If AMD release a CPU at that sort of price I don't think they'd sell many (I don't know that Intel do either) unless it were the second coming of Jesus in CPU form.

However, Intel could take a massive hit on their profits or even a loss if AMD started a price war and still be in good shape. So it's a difficult situation and they will have to find a good balance between profitability and volume of sales and hope that Intel's greed prevents them from reducing prices to rock bottom levels.
 
One of the largest councils in the country... although not in the same league as the ones you mention.

Ah, Council IT is a racket. None of this reflects on you but it has, invariably in my experience been a mess of inside deals, politics and waste. Actual value for money has always been third or fourth on the priority list for any council purchase I've come into contact with. Hell, you probably know this better than I.
 
I'm looking forward to seeing this released, and hoping it's even partly as good as they're saying it is. I'm still on an FX8350, but I chose it because it gave me roughly best bang for buck for my needs (despite it's shortcomings). I game once in a blue moon but I run four, five or six VMs concurrently and do a lot of video encoding. Eight 'cores' and decent multithreaded performance for a hair over £100 on sale was a no brainer compared to the Intel offerings of the time.

Ryzen 8/16 will be a nice replacement if it all pans out how they're hyping it. I'd been looking toward Xeon or i7 chips but Summit Ridge should blow a hole in that for the money. I definitely wouldn't say no to two of their 16/32 chips in a server board, even with the lower clock speed. :o If I gamed a lot I'd possibly change my mind, but even then the SR 8/16 should be more than adequate.

I've been reading this thread and getting bored for some of it, but throughout I couldn't help being reminded my family are traditionally Everton supporters (I'm a geek so I don't care about football lol). Something about living in hope, season after season, that this year will finally be the year you win the Derby/league... Ring any bells AMD? :D Hopefully in 2017 it comes true, because some competition for Intel is badly overdue and I'm happy to throw my money at AMD out of principle if the chips are worth it.
 
As someone who doesn't have the time to read the previous 102 pages, will this be 8 separate cores or modules like crapdozer?

Search is at the top right, and even Google would have told you. It's 8 proper cores with Intel-style SMT giving 16 threads on the top Summit Ridge desktop chip, yes. BTW it's 39 man-sized pages. You're doing it wrong. :p
 
I use a weenie box *cry*

9a93918f846640e1bec6c5e367c0f0fe.png
 
Is there going to be a server chip / platform or is Summit Ridge intended to compete with the E5s as well?

in the background AMD have actually been showing their server cpu work well before desktop zen was shown.

we have a confirmed 32 core / 64 thread chip here which should mean 64 cores 128 threads in a mac pro type chassis

http://www.anandtech.com/show/10581...-naples-32cores-dual-socket-platforms-q2-2017

as most new applications i deal with are going into the cloud now, AMD just need to be in the same ballpark.

what i believe amd will do in the desktop and server space is to price the chips comparatively but to offer at least 2 more physical cores per CPU. e.g. amd 8 core priced at intel 6 core level, 10 amd cores for 8 intel and so on.

the performance will be -+5% per core so it's not really a consideration unless you run a niche app. at the moment the gap in perf for server based apps is between 8-45% in intels favour with many apps in the 25% plus bracket with things like 3dsmax 40%+ this is just enormous.

the major cloud operators (oracle, apple, google, microsoft, ibm) have no vested interest in giving intel more money. what organisations such as local councils do with their internally hosted IT is not relevant as it will be punted into the cloud over the next 10 years.

the important thing is for AMD to start smashing the 8 core and above CPU market as 12 cores per server is the current price perf sweet spot for x86 blades in a 2 cpu per blade config. once you drift from 12 cores per server things start to get hideously expensive.

if i were an investor in intel i would cash in my stock now and see what happens next.
 
Last edited:
Virtualisation makes 12 core*server CPUs cost effective I would have thought. even 20,32 64 cores*would be deemed a good thing for virtualisation ?

Guess it depends what clockspeed/power draw they can hold with 64 cores... Scaling up the 8 core, that could be close to a kilowatt, so that'll probably need cutting down some :D
 
in the background AMD have actually been showing their server cpu work well before desktop zen was shown.

we have a confirmed 32 core / 64 thread chip here which should mean 64 cores 128 threads in a mac pro type chassis

http://www.anandtech.com/show/10581...-naples-32cores-dual-socket-platforms-q2-2017

as most new applications i deal with are going into the cloud now, AMD just need to be in the same ballpark.

what i believe amd will do in the desktop and server space is to price the chips comparatively but to offer at least 2 more physical cores per CPU. e.g. amd 8 core priced at intel 6 core level, 10 amd cores for 8 intel and so on.

the performance will be -+5% per core so it's not really a consideration unless you run a niche app. at the moment the gap in perf for server based apps is between 8-45% in intels favour with many apps in the 25% plus bracket with things like 3dsmax 40%+ this is just enormous.

the major cloud operators (oracle, apple, google, microsoft, ibm) have no vested interest in giving intel more money. what organisations such as local councils do with their internally hosted IT is not relevant as it will be punted into the cloud over the next 10 years.

the important thing is for AMD to start smashing the 8 core and above CPU market as 12 cores per server is the current price perf sweet spot for x86 blades in a 2 cpu per blade config. once you drift from 12 cores per server things start to get hideously expensive.

if i were an investor in intel i would cash in my stock now and see what happens next.

This is quite true, we are in the process of sorting out a new WAN deal, part of this offers direct pipes to these cloud based services, globally as a group we are moving towards more cloud based applications as well.

Locally on site everything we run is off VM now as well, we just phased out our last few physical servers and virtualized the remaining apps they were running.

Market is definitely moving towards this sector.
 
Virtualisation makes 12 core*server CPUs cost effective I would have thought. even 20,32 64 cores*would be deemed a good thing for virtualisation ?

Most licensing is done per core now, which would make it more expensive for the high core count CPUs. Means you have to be more careful while spec'ing the server infrastructure. Microsoft informed us they're switching to a per-core model this year, when we requested a couple more W2k12r2 licenses. Managed to get them added on to our OVS agreement eventually. Would have hurt to license the 48 cores in our VMware cluster.
 
Most licensing is done per core now, which would make it more expensive for the high core count CPUs. Means you have to be more careful while spec'ing the server infrastructure. Microsoft informed us they're switching to a per-core model this year, when we requested a couple more W2k12r2 licenses. Managed to get them added on to our OVS agreement eventually. Would have hurt to license the 48 cores in our VMware cluster.

This is also true, just renewed licensing and upgraded some SQL Enterprise Servers, the cost made me wince, its all per core so we moved them to smaller hosts to lessen the cost lol :)
 
http://wccftech.com/amd-ryzen-zen-cpu-benchmarks-leak/

I guess Ryzen failed to live up to hype after AMD cherry picked bechmarks like Blender and handbrake that showed Ryzen outperformed 6900K.

Now leaks showed Ryzen is slower than 6900K in Cinebench R15 and Fritz Chess but Ryzen is slighter faster than stock 7700K in both benchmarks.

So Ryzen is probably either slower or slight faster than 7700K in games benchmarks.
 
http://wccftech.com/amd-ryzen-zen-cpu-benchmarks-leak/

I guess Ryzen failed to live up to hype after AMD cherry picked bechmarks like Blender and handbrake that showed Ryzen outperformed 6900K.

Now leaks showed Ryzen is slower than 6900K in Cinebench R15 and Fritz Chess but Ryzen is slighter faster than stock 7700K in both benchmarks.

So Ryzen is probably either slower or slight faster than 7700K in games benchmarks.



100% reliable benchmarks when one of the screenshots shows Zen being detected as a 1.0Ghz P3.... wait for Reviews, not random speculation.

edit: retracted - didn't realise the translation of that is "Performance relative to P3 1.0Ghz (480 Kilo nodes per second)"
 
Last edited:
http://wccftech.com/amd-ryzen-zen-cpu-benchmarks-leak/

I guess Ryzen failed to live up to hype after AMD cherry picked bechmarks like Blender and handbrake that showed Ryzen outperformed 6900K.

Now leaks showed Ryzen is slower than 6900K in Cinebench R15 and Fritz Chess but Ryzen is slighter faster than stock 7700K in both benchmarks.

So Ryzen is probably either slower or slight faster than 7700K in games benchmarks.

Or there are quite a few unknowns from this chinese source.

For one windows will need it's task scheduler patched to support Ryzens form of SMT similar to how Intel needed for Hyperthreading. Which AMD would have had in their builds for the Ryzen event. Windows needs to know how to treat the hardware etc.

We also don't know the state of the Engineering sample they tested.
 
Can that crappy website just be banned?
I don't know why people use it.
I'm referring to the wcftech one.
It was a joke when people started using it, it's a joke now.
 
Back
Top Bottom