Monday, September 10th 2018
AMD Announces 2nd Gen Ryzen Quad-core and Energy-Efficient Processor Models
AMD today announced the much-awaited 2nd generation Ryzen quad-core socket AM4 processors, in addition to two new E-series (energy-efficient) variants of its existing processor models. To begin with, the company announced the 4-core/8-thread Ryzen 5 2500X and the 4-core/4-thread Ryzen 3 2300X.
Unlike their predecessors that are carved out of the "Summit Ridge" silicon by disabling 2 cores per compute complex or CCX (2+2 CCX config), the 2500X and 2300X feature a 4+0 config, or an entire CCX in the "Pinnacle Ridge" silicon being disabled. This also means that the 2500X has just 8 MB of L3 cache (its predecessor has 16 MB). The 2300X is clocked at 3.50 GHz with 4.00 GHz boost, while the 2500X ticks at 3.60 GHz with 4.00 GHz boost. The TDP of both chips is rated at 65W.
AMD also released the "E" brand extension for its 2nd generation Ryzen series, with the new Ryzen 5 2600E, and the Ryzen 7 2700E. Both these chips sacrifice clock speeds for an impressive 45W TDP. The 2600E is clocked at 3.10 GHz, with 4.00 GHz (compared to 3.60 GHz ~ 4.20 GHz of the 2600X); while the 2700E ticks at 2.80 GHz, with 4.00 GHz boost (compared to 3.70 GHz ~ 4.30 GHz of the 2700X). The company didn't reveal pricing of the four chips.
Source:
Anandtech
Unlike their predecessors that are carved out of the "Summit Ridge" silicon by disabling 2 cores per compute complex or CCX (2+2 CCX config), the 2500X and 2300X feature a 4+0 config, or an entire CCX in the "Pinnacle Ridge" silicon being disabled. This also means that the 2500X has just 8 MB of L3 cache (its predecessor has 16 MB). The 2300X is clocked at 3.50 GHz with 4.00 GHz boost, while the 2500X ticks at 3.60 GHz with 4.00 GHz boost. The TDP of both chips is rated at 65W.
AMD also released the "E" brand extension for its 2nd generation Ryzen series, with the new Ryzen 5 2600E, and the Ryzen 7 2700E. Both these chips sacrifice clock speeds for an impressive 45W TDP. The 2600E is clocked at 3.10 GHz, with 4.00 GHz (compared to 3.60 GHz ~ 4.20 GHz of the 2600X); while the 2700E ticks at 2.80 GHz, with 4.00 GHz boost (compared to 3.70 GHz ~ 4.30 GHz of the 2700X). The company didn't reveal pricing of the four chips.
89 Comments on AMD Announces 2nd Gen Ryzen Quad-core and Energy-Efficient Processor Models
I imagine it can be a render beast, a cheap server, or pair it with a quadro and do CUDA work without needing a high TDP cpu.
Don't look at it as a desktop CPU, look at it as a server / work 8 core CPU with a desktop 's price.
let's take a 4c4t chip ( as you mentioned) running at 4 ghz having 95 w tdp
let's give this chip a "significant load" which almost use all cores up to 100% and reaching max tdp but no throttle
now let's give same load to the 8c16t chip, running at 2,8 ghz having 45w tdp ; will this throttle?
you say yes but i say now if the work of a 4 core working 100% is sent to 8 core than overall load is less than 100% even having a lower freqv. and within tdp so no throttle
the above is not feasible only if the "significant load" can't use all available cores...
Maybe google inlet 4XXX S and L ....
Most thin clients are really low powered CPU's
And yes a Igpu would be nice on a low powered version/.
The only reason to make a low-power CPU drawing 35-45W (vs "normal" 65W) is to meet cooling limits of USFF cases (especially those passively cooled).
But this CPU has no IGP, so it's not suitable for existing USFF systems.
Maybe we don't know something. Maybe AMD has an awesome deal with OEMs, who will make mini PCs with GPU on mobo. This could happen.
But wouldn't it be better for AMD to - once in a while - make a product that meets existing needs, rather than betting on an incoming revolution? :-)Just imagine it having single-core performance of... ah... OK... you have an FX...OEM PCs are as simple as possible, no nonsense products.So let me understand this better. You really think that people who buy off-the-shelf PCs use them primarily for Internet and games?*
Does this also apply to laptops?
*) isn't this a fairly accurate description of many TPU forum members? :-)
AMD has a very strong position in thin clients thanks to low-power APUs.
But when you move to the 35W+ range (mini PC, SFF and so on) it's basically Intel's full domination.
AMD hopes that PRO 2200GE will become a serious contender. We'll see how it goes...
My idea of a "significant load" is threads that will push a core to 100%. Games, web browsing, office stuff doesn't multithread indefinitely and are better served by faster cores. Since I'm not buying cores by the pound (any more than I bought GHz by the pound back in P4 days), I tend to pick fewer cores that can actually sustain higher frequencies. This is also part of why I hate laptops and their throttling CPUs, but that's another story.
Basically I was thinking of a few CPU intensive threads, while you were thinking about load made up of many lighter loaded threads. I believe that was our disconnect.
Remember it's just TDP. You CPU uses a lot less most of the time. And when it get near (or past...) TDP in tasks like gaming, it's because it needs that power.
You, precisely, can just buy a normal Ryzen 7 and downclock it to lower power consumption. Every consumer can. AMD could even bundle an app that lets you quickly move between TDP levels.
35W - 45W exist to meet USFF cooling capabilities. So that a large Excel spreadsheet or SAS program doesn't fry the system. :-)
And the other nice thing is that a system built around a 35W CPU can easily be squeezed under 65W and, as a result, it can be powered with an ordinary 65W notebook charger (ubiquitous in offices).
Mmm, 35w 6 core APU. One can dream.
We're getting more cores simply because Intel and AMD hit a performance wall and suddenly making more cores was cheaper than making faster ones.
Once you move past 2 cores (which profoundly improves PC use comfort), more cores is always worse, because there's more potential wasted on just supporting the architecture.
If your life expectancy at this point is beyond 30 years, it's very likely you'll see 2/4-core CPUs' comeback. Also in HPC (if not especially there). :-)
At least 2 of my customers can benefit from it. One is a webdev company, who's PCs usually include a low-end dGPU solely for multi-display functionality (you may not believe it, but it's nearly impossible to find a motherboard with 3 digital display outputs in 2018). Having an iGPU just to have it is not a good justification.That's nonsense. You have a multithreaded OS with a bunch of multithreaded software, you have multithreaded services and multithreaded games, and you are saying that more cores is a wasted potential? Even if you have the fastest-ever dual-core, you ain't gonna get the same multithreaded performance as a quad-core with half the clock speed, because dual-core will waste tons of time on context switching (as a "tech savvy" user you are probably aware of this).
And what does it mean "supporting the architecture", in regards to cores? Is it harder to support a quad-core CPU rather than dual-core? You lost me there...
We've been running hundreds of threads for years, that does not automatically translate into a need for more cores.
I'm sure someone's gonna test it right away.