Reviews AMD 2nd Gen Ryzen 7 2700X and Ryzen 5 2600X Review

Rendering / encoding have been developed to take advantage of more cores since they require performance far in excess of what one core could ever provide.
Gaming is in the absolute majority of cases GPU limited.

But how much of the everything else stuff is left once you exclude all the stuff so old (or non demanding) that it runs fine on a potato (potato = FM2 APU, mobile Celeron, Atom etc)?

As I said: "actually limited by the CPU", no one cares if Word runs one nano second faster with better IPC, because even running on an Atom it can already deal with more than the meat sack at the keyboard.

 

Luckily in any scenario where said potato is sufficient it being a potato has no relevance.

 

In the days of dual core i3s that was indeed a good argument, but now with 6 cores for less than £200 from both Intel and AMD? Not so much any more.
(unless of course you try something like doing an h.265 encode of a blu ray while playing Cities Skyline, then you better have a Threadripper).

 

The way operating systems schedule processes and threads is in that black box over there, whenever I've looked inside it seems to go on forever, a bit like the tardis.

 

Not unless your cpu was maxed out by the game which most aren't. It's not like 1 core ever just does one thing (e.g. the game main thread) - the reality is they switch what they are doing many times a second so any single thread will be able to stay maxed out at a full core's worth of time unless your cpu is completely swamped. My experience of background processes is that it's not the cpu they kill but the disk and internet access. They are much easier to max out. That and the really irritating ones that force you back to windows mid game to show you some stupid dlg saying they have an update.

Back on topic CPU looks really good, it feels like they are only 1 step behind intel now on IPC while at least equal on power consumption and ahead on core count. Considering where they were a few years ago that's amazing. Reminds me of the A64 days - then AMD went all short term failing to innovate and just suck up as much money as they could, while Intel's Israeli's developed the core architecture. Now it feels like it's Intel not going anywhere fast, if AMD can keep their rate of development up they might actually overtake Intel in IPC.

 

Had an 8 core for 3 years. Never felt I'm even getting close to the edges until I (rarely) do some encoding.

 

Remember when the first generation of decent (i.e. not the JMicron controllers) consumer SSDs became available, and suddenly everyone realised the massive benefits to responsiveness they gave due to completing operations faster (low latency), even if raw transfer rate was no higher (and often lower for sequential access) than a HDD?
Same with single-threaded tasks: even if a task cannot max a single core, completing that task faster makes a machine more responsive, which is very noticeable in everyday tasks. Reducing the time it takes to do a task has an immediate impact on responsiveness. A faster core will always directly translate into a faster task completion time regardless of parallelism, while adding more cores will only benefit parallelisable tasks.

Then there are workloads that are latency-sensitive rather than throughput sensitive. Examples are real-time audio editing, and VR and AR. Currently these are as much a subset of workloads as highly threaded non-GPU workloads are, but VR and AR specifically are workloads that are going to grow more common, and are already driving design changes to GPU architectures.

 

This, cast your mind back to when dual-core CPUs were brand new, they were insanely responsive not because there were a lot of threaded applications but because workloads were split. Games have been moving in the same direction slowly and they will continue to become more multi-threaded.

 

You know that ryzen came out and is not far off Intel now (it's only 1Ghz that is keeping single threaded behind, clock both at 4ghz and they have around same single threaded performance) ryzen is no faildozer (all fx and type A cpus) where single threaded performance was about half of Currant and most past Intel and ryzen cpus

Zen3 may close the gap on Ghz but that's long away

I may get 2700x to replace the 1700 I have as I someone wants a new pc so all I need is cheap b motherboard and 2x8 16gb ram plus my cpu, if xfr2 and pb2 did not work on the same x370 I have I would not bothered to think of doing it, (at the time I thought that x470 was a requirement for pb2 but AMD is not Intel it seems)

 

I’m really not a fan of this XFR2 thing on my X370 CHVI. It thinks that 1.59v is fine... I know it’s temperature related to a degree, but that’s way too much for my liking.

At least it does boost to 4.35, so we shall see what it can do on a manual OC.

 

Looks like Store MI is actually quite useful for people who can't afford big SSDs.

 

This is why MS' desicion to limit Tiered Storage Storage Spaces to enterprise versions of windows [sure you can do it on pro and possibly even home iirc, but it's a ballache to setup as it requires some command line trawling] annoys me. Sure it's not for everyone, but it something that there are uses for outside of enterprise setups.

 

Well you can have it on your ryzen system for free though limited to a 256Gb fast cache (SSD) and 2Gb ultra fast (RAM) cache, still a great option for people who want to accelerate a big spinner. X470 and X399 supported.

 

When i read that it seemed nuts, not nuts as in doubting you but in that it would go that high, but after having done some research it seems it's normal, AMD_Robert posted on reddit saying that "Temporary blips into the range of 1.4V and 1.5V are completely normal, part of the design, and totally harmless"

 

I could have for free now, if i had the requiside hardware and could be arsed with the ballache of setting it up [and don't need to boot from the tiered pool]... My gripe is more that i shouldn't need bundled software to allow me to do what windows can [technically] already do.

 

Well, if they’re totally normal, then I’ve got no issue running it at 1.46v daily. It’ll only be in the rig for a year until the next iteration anyway.

 

He does go on to say those high voltages are short spikes and that average 24/7 voltages should be around the 1.1v 1.2v range, a constant 1.46v core voltage seems rather high .