AMD Launches 7th Generation Bristol Ridge PRO APUs

  • Thread starter Thread starter Guest
  • Start date Start date
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
The marketing slides are lying about the relative x86 performance of A12-9800 compared to i5-6500 - independent benchmarks are going to show that i5-6500 generally executes more x86 instructions per second than A12-9800.
 
"AMD provides 18 months of image stability" What does that even mean?
 
If the OEMs pack the machines with crippled performance motherboards like the first HP AM4 motherboard the performance will suck even worse. I have worked on three recent AMD laptops that were fit only for a landfill because they were restricted to single channel memory, no matter how much memory you put in. I benchmarked all three machines and they all suffered more than a 50% performance penalty from their advertised specs because the single channel memory. My father-in-law's laptop with an A8 was unable to open MS Word faster than a 3.2 Ghz P4-HT Dell Desktop (Windows 7 vs Windows 10 so not exact comparison) same father-in-law has. He thought there was something wrong because his "new" $350 laptop could not keep up with the old desktop I gave him. I could not believe he was so stupid as to buy without asking me first, but whatever, he got used to the lag. And it lags bad, as in click, wait 5-6 seconds, and then something happens, then click again. I would pull my hair out if I had to work on that.
 


Basically, it means that the components and drivers won't change over the defined period of time. This minimizes IT overhead for qualifying/deploying new images, etc. So the business makes images of systems, then just flings them around wherever needed when they deploy new systems. However, if there are driver/component changes an image won't work on another system.

"A consistent and key condition for the commercial market is disk image stability which allows IT organizations to focus on supporting a single, stable platform, reducing the number of disk images, and in turn, helping to reduce overall IT support costs. With disk image stability, businesses can create and maintain a standardized software image consisting of hardware drivers, application software and an operating system. This software image can then be applied when new PCs are deployed without the concern of incompatibilities between hardware and software." -- http://www.businesswire.com/news/home/20050926006172/en/AMD-Introduces-Commercial-Stable-Image-Platform-Program
 
AMD releasing subpar garbage since 2009. what a joke. i'll stay with intel all day.

come on AMD show me some high end stuff!
 
All of that stuff crammed on the A12 while stays at 35watt TDP package impress me. I remember HD7750 that also has the same number of SP hovers at 60-65C.

Alas, its only on newer platform. I would want it to exist at FM2+ too.
 


Actually, Excavator is pretty decent and not that old. It is a far cry from the other Bulldozer variants regarding performance per Hz and power efficiency. While it would've been nice to see Zen, Excavator with a platform upgrade is a good way of getting AM4 started. The more things you have already tested when you bring out Zen, the fewer things that are likely to go wrong.

 
Agree - Kaby Lake will likely be the main performance winner. But ZEN 6+ cores may find some good uses depending on performance results.
 
HP will cripple the crap out of these and put the slowest memory available in them. I bet most will come with one stick of 8GB DDR4 CL15 2133MHz. If there was a way to put slower 1600MHz CL11 ram in they would.
 


Happens all the time in the OEM world unfortunately. The issue that was affecting AMD recently was that some budget solutions were built to the lowest common denominator and lacked even the capability to be upgraded to dual-channel, regardless of the number of memory slots. This was a big issue with early-gen FP4 platforms where a cheap common board was designed for Carrizo-L but was also used for a lot of OEM Carrizo laptops. Higher-end mobile models tended to be safe but were also rare and priced a bit too high.

From what I'm hearing this shouldn't be a common occurrence with the newer FP4 platforms, as long as it has at least two memory slots. The AM4 desktops in particular being released should be fine. Even if the OEMs choose to be stupid and use poor memory configurations, anybody who actually knows/cares would be able to easily upgrade it themselves. Also is anyone else stunned by the huge clock gains they've managed to pull off on the mobile side with Carrizo vs. Bristol Ridge on the same 28nm process? In particular look at the base clocks. Big jumps for most model replacements.
 


Link? I'm a little skeptical as to how easily it hit that number, because if it was a great overclocker I suspect they'd have some K variants. If this was a "typical" result on aircooling I'd be impressed. Otherwise, we've done it with Godavari already.
 


In what way is the Excavator "decent"? When I googled reviews, Intel was 100%-300% faster in encoding, 50%-100% faster in synthetic benchmarks, and 30%-100% faster in actual games. The only thing the Excavator competed in was power, it only consumed about 10% more than Intel, but was 1/2 the performance on average.
 


Good eye! A bit of late night typo-ing, fixed. :)

 
Happens all the time in the OEM world unfortunately. The issue that was affecting AMD recently was that some budget solutions were built to the lowest common denominator and lacked even the capability to be upgraded to dual-channel, regardless of the number of memory slots. This was a big issue with early-gen FP4 platforms where a cheap common board was designed for Carrizo-L but was also used for a lot of OEM Carrizo laptops. Higher-end mobile models tended to be safe but were also rare and priced a bit too high.

From what I'm hearing this shouldn't be a common occurrence with the newer FP4 platforms, as long as it has at least two memory slots. The AM4 desktops in particular being released should be fine. Even if the OEMs choose to be stupid and use poor memory configurations, anybody who actually knows/cares would be able to easily upgrade it themselves. Also is anyone else stunned by the huge clock gains they've managed to pull off on the mobile side with Carrizo vs. Bristol Ridge on the same 28nm process? In particular look at the base clocks. Big jumps for most model replacements.
Click to expand...

First, the same thing can happen again. The FX-9830P for example is configurable from 25-35 watts. And yes, 10 watts can make a big difference when trying to squeak the most performance out of it. Secondly using 1 stick of ram is just crippling the processor out of the gate, which doesn't need any help in that category. And lastly even if they've always done it doesn't make it right or acceptable.
 


This idea was not new. One solution would be applying HBM memory as GPU's own memory pool. Which still can be shared to CPU should AMD desire as L4 cache or something.
Considering current CPU architecture performance AMD has, its plausible they'd want to wait on how Zen fares. Complementing Zen's potential much further, along with latest fabrication node for the GPU part.

Then again, we can only hope they'd eventually going on that direction. Hopefully soon.
 

It sounds like you're basically describing L3 cache?
 


I've seen Excavator more like 70% to 80% of Haswell and Skylake in per core performance at the same frequency, which is a lot better than the roughly 50% of their Steamroller and Piledriver cores compared to Haswell and Skylake. Are you sure you aren't looking at Steamroller or Piledriver benchmarks? Can you show me your links?
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom