With 400 Series gaining Inventory,

Etailers are clearancing the 300 Series,

Seeing 380s for $140, and 380Xs for 179.

Intel plans a 10nm node next. Some folks are speculating that Intel 7nm will be delayed to 2022.

Article

As noted in an above post, others hypothesize (based on limited information) that AMD will skip the 10nm node entirely and go straight to 7nm. That makes sense because 10nm may not be that much faster than 14nm -- and some folks (like gamers) just want speed.

Some think 7nm will be the end of the line for downsizing silicon circuits -- and may be hard to do. If AMD succeeds at 7nm first, their CPUs might surpass Intel (for the second time this century).

This is speculation and guessing in 2016. But, it does show that big increases in CPU and GPU performance are less likely in the near term. Thus, our current upgrades to 14/16nm CPU and GPU may last a few years as the 10nm node may not be "significantly" faster and may be later than planned -- and 7nm may be much later than originally expected -- these are upgrade cost/benefit considerations for those of us AMD fans upgrading early next year (Zen CPU and Vega GPU).

These are 28nm AMD APUs. I've built a couple computers using the older model APU. Based on my limited experience, SimHQ folks should not be looking to an APU (from AMD or Intel) for an acceptable PC gaming experience. Don't buy an APU intending to match it with a separate GPU. Just get a "stand alone" CPU and "stand alone" GPU.

However, for those on a really tight budget, the new APU should beat Intel for low cost (without a separate GPU) gaming in a low cost PC or laptop -- because they do have excellent AMD GPU circuitry (Intel's in-CPU-chip graphics are improving but still second rate). Based on my experience, you can play all the games with the APU -- but, not to SimHQ standards

These will use the new AM4 motherboard technology. The AM4 motherboards will run ALL of the upcoming AMD CPUs and APUs.

I Steam VR benchmarked my current RX480 powered system (listed below). Mostly, frame rates were above 100fps and never below 90fps.

My RX480 system just meets the full standard (does not exceed it) as shown by the screenshot below. That is what AMD advertised -- they said the RX480 would fully meet the VR specifications but not exceed. Parenthetically, the FX9590 CPU also fully met the specification.

During the run, two images are up on the screen -- side by side. I let my "eyes" "spread" (opposite of cross) to watch the scene in full 3D stereo without needing special glasses or a VR setup. I suppose others could do the same.

It took me both of my 7970 Lightnings to Get to the Green, lol.

This is an upgrade of current Playstation 4 with substantially improved graphics. It it is said to use the latest AMD graphics (as found in RX480).

The new Microsoft "XBox One S" also uses an upgraded AMD chip.

As noted other times in this long thread, this implies that new console games will be optimized for AMD graphics. Of course, many of those games will become PC games also -- some with Nvidia tweaks -- but, underneath, primarily AMD graphics.

seeing as Xbox One S is sporting "5 TFLOPS" I'd Say it's a RX480 GPU embedded.

both AMD and Sony have been spamming my inbox with "Polaris on Consoles" Emails/Tweets, lol.

These are a hotfix. Not a new WHQL set.



Rumor is, AMD will bypass 10nm node and go straight to 7nm. We'll see.

A DX12 benchmark set that looks good for AMD. The RX480 is even competitive with Nvidia 1070 in this game in DX12 at 1080p.

As previously noted, AMD does better than most folks expect when using the most up to date graphics APIs (DX12 and Vulcan). So, AMD RX480 "bang for buck" makes sense if one has to keep a card for a few years -- as DX12 and Vulcan titles are released.



Article

I will never, NEVER, play a game without some form of FSAA ever again. Jagged lines are the ultimate experience destroyer to me. The whole screen looks like utter filth.

No-FSAA benchmarks are up there with comparisons at 640x480 resolution for me. How does a card perform when it's set the way I would actually play a game? That's all that matters. If a game doesn't have FSAA, I force it thru drivers every time. You can make a 15 year old game surprisingly decent despite low res textures with high FSAA.



The Jedi Master

I wondered if the no FSAA in the tests might bother some folks

Personally, I never use FSAA (one exception where I turned it on by default when selecting Ultimate settings for a game).

Historically, once we got to 1080p, I always set AA to 0 -- makes the game "blurry" to my eyes on a "good eyes" day (normal monitor at about 30 inches view distance). Occasionally 20/15 vision at computer distance on a very, very "good eyes" day. On the other hand, dry eye can somewhat cloud things at other times -- so AA is unnecessary

It has occurred to me that others may see things clearer and have faster eye response than I have. So, they may prefer AA and higher frame rates than are okay for me. Thus, from time to time, I mention that I use 0AA and mention that 30fps looks smooth to me -- so folks understand where I'm coming from when I say something "looks good" and "plays smooth".

In summary, yes, I agree. Giving the FPS both with and without AA seems relevant.

I actually do wonder if the different forms of AA now available and the different AA algorithms used by AMD and Nvidia makes a fair/true "1 to 1" comparison more difficult -- so they just skip AA at times.

The DX12 update on DE:MD has been a disappointment for me.

The internal benchmark of the game says I gained 8% on average FPS (~56), but the whole thing is a stuttery mess on the actual game.

On DX11, on a mix of Ultra/Very High settings, runs very smooth (51 Average) and looks gorgeous.

Even on a R9 390X with 8GB, it only reverts to smooth on DX12 with default High settings. My Average FPS goes up to 65.

And yes, I'm using Crimson 6.9.1.

Yes, the raw fps rate is no longer a sufficient barometer either. If your fps averages 60 by spending half its time at 120 and half at zero, that's not good.

Stutters, dropped frames, tearing, hitching, there are various issues that don't appear in a given bennchmark but may make a game unplayable for some users. That's why people who say "it has to be 60" make me laugh because in a vacuum that's meaningless. Give me a smooth issue-free low frame rate over a "fast" rate with all that junk in it.

There were some games in the past with the ability to have dynamic details, it would crank it up when the game ran fast and smooth and start lowering it when the game slowed down. This has fallen out of favor, although with most games being lazy console ports the performance is usually fine anyway.



The Jedi Master

Article

Intel will have 10nm quite soon. But, is putting 7nm off until 2022, it is rumored.

AMD has multiple partners making "silicon" for them. Each approaches things a bit differently.

Recent announcements, rumors, speculations suggest 12nm, 10nm, or 7nm as AMD's next major process node (after 14nm). Interesting to me, is the timing of all the new nodes.

The other thing to notice is that actual "speed" increases relatively little (15 percent) but power efficiency is way up.

Competition is good

Article

In connection with my above post: A long technical article that discusses how all "nodes" are "not created equal" and other technical stuff.

The summary conclusion is that Intel does nodes best. And Intel's 7nm will be the first "true 7nm" -- even if it is the "last 7nm" on the market.


Still, the actual performance differences between nodes are not real big in terms of making a game FPS difference people can "see" and "feel" (versus only "measure") -- and games are the toughest test for most PC buyers. Thus, being last with the best 7nm may still equate to being last in the games arena -- or not. We'll see. Of course, I expect Intel to stay "at the top" in the CPU world in sales volume -- no matter what they do

See we AMD fans at SimHQ can "take __it" and smile

Anyhow, the article is probably correct. AMD will be one node of experience behind Intel. As the Intel 14nm is a step in their architecture path but is all new to AMD (so AMDs version may not be "perfect").

However, that may only translate to about a 5 to 15 percent per core-clock advantage for Intel. Far less than the advantage has been recently. And, I would claim that 15 percent in a CPU will not be truly noticeable in most uses (in the office or home).

Moreover, the AMD 8 core / 16 thread processors may be selling for enough less than Intel that AMD would be the "rational choice" for most uses. Although, I expect most would buy Intel due to previous purchases and brand recognition.

However, we'll see. Nothing may turn out like I expect. It would not be the first time (I still remember "Bulldozer")

Competition is good

is this the same analyst that swears AMD is lying about 40% more than Steam Roller and says it's more like 40% over the Core that Came before Bulldozer (Orachi I think)

Even this "biased" analyst agreed that the new AMD Zen actually is 2 percent faster (per clock) than the Intel chip it was tested against -- in the test that was run.

In effect, he's just saying the new Intel chip will be faster than the old Intel chip (to which I say -- no kidding). Moreover, he is "guessing" that AMD Zen will not improve between engineering and production (could be true).

And then, he writes an article -- as he is paid to do

Then, a writer known for his "smirky" style writes an article about the analyst's article -- as he is paid to do

Finally, I copy the smirky words over to SimHQ -- for the fun of it

As we all know, we'll find out the truth when the products are on the shelf and independently tested -- prior to that is rumor and guesswork (sometimes educated guesswork).