via has a x86 license for the time being...can't really sue them for that. Besides AMD's puma+ cores are better than VIA's isaiah+ cores. Even wit high clock, unless they have been radically redesigned, it aint gonna do much trashing.
via has a x86 license for the time being...can't really sue them for that. Besides AMD's puma+ cores are better than VIA's isaiah+ cores. Even wit high clock, unless they have been radically redesigned, it aint gonna do much trashing.
Carrizo and Kaveri were not so good updates from Bulldozer. Also their Carrizo Athlon seems inferior than the previous two generations.
DrMrLordX
Lifer
- Apr 27, 2000
- 21,634
- 10,850
- 136
I agree that it will be interesting. It will also let us compare Carrizo (x4 845) and Bristol Ridge with DDR3 vs DDR4 on what is essentially the same memory controller. Though the per-clock improvements of Carrizo over Kaveri can be as low as 5%, which is evident from The Stilt's testing of Cinebench R10 on his Carrizo reference platform.
Yeah, half the cache and half the pcie lanes. Killed any appeal it might have had for me
I'm not sure what to make of the PCIe lane situation. But yeah, it's quite a limitation considering the fact that you have to use a dGPU with the x4 845. If you want any other PCIe devices in the system . . .
Also, it may offer us a preview of what we can expect from Bristol Ridge on AM4 wrt PCIe lanes. The Summit Ridge/AM4 slide we've all seen shows that the 16x lane PCIe configuration is particular only to Summit Ridge. Bristol Ridge may have an identical configuration to Carrizo.
No problem. Though Stilt offered more complete information of the cache differences, so . . . yeah. I can only claim so much credit.
When overclocked, many 860k processors are going to be faster than the x4 845. The fastest the 845 will ever go is 4 GHz (barring bclk OC, which sucks on FM2+ for the most part). I would expect a 4.5 GHz 860k to outperform the 845 in most situations (with a few rare exceptions).
I certainly would like that. It would also be nice if they provided a way to just run all zones (NB, CPU, iGPU) at full tilt per user configuration assuming everything can stay within an acceptable thermal envelope. Kaveri offered no real way to do that from the UEFI which was irritating. Sure, it SAID you'd get static clocks, but in reality . . .
Last edited:
AtenRa
Lifer
- Feb 2, 2009
- 14,001
- 3,357
- 136
I dont like that it only has 8x lanes but 8x PCIe Gen 3.0 has the same bandwidth as 16x Gen 2.0. So for a single GPU its fine, you will only loose 2-3% on a High-End GPU. The majority of the users that will buy a BristolRidge will not pair it with a $500-700 GPU anyway.
But again it seams that BristolRidge is the same die as Carrizo. If it cannot use more than 2400MHz DDR-4 memory then its not going to bring anything interesting in to the desktop except lower consumption and the new FM4 boards. In fact the FM4 will be the selling point and not BristolRidge, since you will be able to upgrade to a ZEN CPU or ZEN APU next year.
But again it seams that BristolRidge is the same die as Carrizo. If it cannot use more than 2400MHz DDR-4 memory then its not going to bring anything interesting in to the desktop except lower consumption and the new FM4 boards. In fact the FM4 will be the selling point and not BristolRidge, since you will be able to upgrade to a ZEN CPU or ZEN APU next year.
The Stilt
Golden Member
- Dec 5, 2015
- 1,709
- 3,057
- 106
All of the 15h parts are limited to 2400MHz max. by the design. The "new" memory controller introduced in Steamroller is the worst piece of hack job I´ve ever seen. It is some off-the-shelf solution "integrated" (whisky tangoed) to the design by using some duct tape and hot snot. It´s functionality relies fully on vast amount of C code (AGESA).
The memory controller in Excavator designs is basically the same thing, the only major difference being the memory types it supports (DDR3/GDDR5 combo vs. DDR3/DDR4 combo).
The "quality" of the memory controller matches pretty closely the rest of the Steamroller design: nothing works (together). Steamroller is like the last ditch design with all of the scrap thrown into a same can and slightly stirred. Each of the different parts work well separately, but when combined together the outcome is just absolute junk. Normally this type of bugs and errata can be found in early evaluation silicon, not in something that gets released to the market.
I certainly hope (pray) that anyone who worked on Steamroller have had nothing to with the upcoming 17h family (or any other AMD product for that matter) :'(
The memory controller in Excavator designs is basically the same thing, the only major difference being the memory types it supports (DDR3/GDDR5 combo vs. DDR3/DDR4 combo).
The "quality" of the memory controller matches pretty closely the rest of the Steamroller design: nothing works (together). Steamroller is like the last ditch design with all of the scrap thrown into a same can and slightly stirred. Each of the different parts work well separately, but when combined together the outcome is just absolute junk. Normally this type of bugs and errata can be found in early evaluation silicon, not in something that gets released to the market.
I certainly hope (pray) that anyone who worked on Steamroller have had nothing to with the upcoming 17h family (or any other AMD product for that matter) :'(
What's your 'inferior' mean? You compare an Athlon X4 to FX octal, right? If so then I agree it's inferior.
Do you have any estimates on how much performance Steamroller have lost due to this?
The Stilt
Golden Member
- Dec 5, 2015
- 1,709
- 3,057
- 106
If we expect that a proper & properly implemented memory controller would reach similar frequencies (2666MHz+) that Intel IMCs do, then the performance penalty for the high-end Steamroller APUs in 3D is around 15%.
The core µarch is what it is and a proper memory controller would not make it any better in general. There are some cases where a proper IMC would improve the performance quite significantly (e.g data de/compression, SuperPI) which is latency & bandwidth critical. The memory latency on Steamroller is over 30% higher than on the previous generation at the same settings (MEMCLK and timings).
If even The Stilt rants about how terrible Steamroller is, it just gives me chills that what stands between AMD going bankrupt and Intel total dominance on the PC market, is that Zen becomes a critical success. We need like three or four miracles in a row, a single one is not enough to make that happen...
DrMrLordX
Lifer
- Apr 27, 2000
- 21,634
- 10,850
- 136
Why do you think AMD hired on Jim Keller? The Kaveri memory controller was cribbed from Altera I think? No matter how good or bad was SR, that memory controller was apparently not designed in-house.
Dresdenboy
Golden Member
I'm sure, the differences we se can be attributed to the different prefetchers and some additional latencies in the ways of requests and data delivery to the cores.
The Stilt
Golden Member
- Dec 5, 2015
- 1,709
- 3,057
- 106
Excavators (CZ, BR) use almost identical memory controller IP as Steamroller does. The only difference is that Excavator has DDR4 Phy in place of GDDR5 Phy found in Steamroller. The DDR3 Phy is identical.
It is bat*hit crazy that the new memory controller has that much HIGHER latency. At 2133MHz CL12-12-12-2T (3400MHz CPU, 1200MHz NCLK) AIDA64 (5.60.3755) measures 112.4ns latency for the DRAM on Carrizo. Richland running with identical settings (excl. 1500MHz NCLK) gets 70.8ns.
Steamroller, which was the first design to implement the completely new memory controller has over 30% higher latency than Piledriver based Richland. For some reason the latency has increased even further in Excavator. While AIDA64 isn´t exactly the optimal method to measure the DRAM latency, based on real world applications (e.g. LZMA, RAR de/compression) it is clear that the memory latency is infact significantly higher than on the older memory controller. Quite an achievement, since the old memory controller was introduced in K8. Until it´s decommissioning it received only minor upgrades.
It is bat*hit crazy that the new memory controller has that much HIGHER latency. At 2133MHz CL12-12-12-2T (3400MHz CPU, 1200MHz NCLK) AIDA64 (5.60.3755) measures 112.4ns latency for the DRAM on Carrizo. Richland running with identical settings (excl. 1500MHz NCLK) gets 70.8ns.
Steamroller, which was the first design to implement the completely new memory controller has over 30% higher latency than Piledriver based Richland. For some reason the latency has increased even further in Excavator. While AIDA64 isn´t exactly the optimal method to measure the DRAM latency, based on real world applications (e.g. LZMA, RAR de/compression) it is clear that the memory latency is infact significantly higher than on the older memory controller. Quite an achievement, since the old memory controller was introduced in K8. Until it´s decommissioning it received only minor upgrades.
AtenRa
Lifer
- Feb 2, 2009
- 14,001
- 3,357
- 136
Yea i have noticed the increase in memory latency from Trinity to Kaveri in my APU wars review.
Llano Memory latency at 65.0ns
Trinity (PileDriver) memory latency at 63ns
Kaveri (Steamroller) Memory latency jumps to 80ns. It will raise higher if you use lower clocked memory like 1866MHz.
Llano Memory latency at 65.0ns
Trinity (PileDriver) memory latency at 63ns
Kaveri (Steamroller) Memory latency jumps to 80ns. It will raise higher if you use lower clocked memory like 1866MHz.
FlanK3r
Senior member
- Sep 15, 2009
- 312
- 37
- 91
Stilt, but its possible to get some pieces very high at IMC:
http://forum.hwbot.org/showpost.php?p=429691&postcount=94
This guy benched 3D Marks with 2800+ MHz memory effective clock on air
http://forum.hwbot.org/showpost.php?p=429691&postcount=94
This guy benched 3D Marks with 2800+ MHz memory effective clock on air
The Stilt
Golden Member
- Dec 5, 2015
- 1,709
- 3,057
- 106
Was he running at default VDDNB levels (~1.25V)? :sneaky:
The memory controller in Steamroller can rarely do anything higher than ~2450MHz MEMCLK 24/7 at sane voltages. The memory cannot be trained even partitially for anything higher than 2400MHz and fully for 2133MHz. It won´t pass validation even at 2400MHz and that´s the reason why 2400MHz was never officially available on DDR3.
DrMrLordX
Lifer
- Apr 27, 2000
- 21,634
- 10,850
- 136
I've gotten DDR3-2640 stable before. The downside was that my NB stability went to hell. I can run NB @ 2100 MHz with DDR3-2400, or 1900 MHz with DDR3-2640 PLUS all the headaches from bclk overclocks (basically, I needed a PCIe SATA controller, and it messed with my iGPU in some weird ways). I lost performance in some areas and gained it in others.
The only way I see Bristol Ridge making any hay with memory overclocks, is if a). AMD implements bus locks for all bclk-related systems EXCEPT memory/NB/iGPU, or if AMD implements a bus lock for only the 4x PCIe link to Promontory/Ponderosa, allowing the end-user to use the SATA ports controlled by Promontory/Ponderosa instead of the ones linked directly to the APU. That would also require no PCIe cards beyond bclks of . . . maybe 120 mhz (my limit with a cheap PCIe adapter on FM2+ was 119 mhz, and that was still sketchy as hell). If they bus-locked all the PCIe lanes then of course PCIe would not be affected.
If they don't bus lock anything, it might still be workable if we can increase voltage to Promontory/Ponderosa and/or the 4x link feeding it.
The only way I see Bristol Ridge making any hay with memory overclocks, is if a). AMD implements bus locks for all bclk-related systems EXCEPT memory/NB/iGPU, or if AMD implements a bus lock for only the 4x PCIe link to Promontory/Ponderosa, allowing the end-user to use the SATA ports controlled by Promontory/Ponderosa instead of the ones linked directly to the APU. That would also require no PCIe cards beyond bclks of . . . maybe 120 mhz (my limit with a cheap PCIe adapter on FM2+ was 119 mhz, and that was still sketchy as hell). If they bus-locked all the PCIe lanes then of course PCIe would not be affected.
If they don't bus lock anything, it might still be workable if we can increase voltage to Promontory/Ponderosa and/or the 4x link feeding it.
http://shop.lenovo.com/us/en/laptops/ideapad/500-series/500-15-inch-amd/?sb=:000001C9:0001625F:
Ideapad 500-15 - 80K40016US
Anyone know if the discrete R5 M330 will crossfire with the integrated gpu? If not, this brings up a interesting question. What is faster the integrated gpu in the 8800p or the discrete M330?
Ideapad 500-15 - 80K40016US
Anyone know if the discrete R5 M330 will crossfire with the integrated gpu? If not, this brings up a interesting question. What is faster the integrated gpu in the 8800p or the discrete M330?
burninatortech4
Senior member
- Jan 29, 2014
- 671
- 381
- 136
The 8800p is much faster. My a10-7300 lappy has an r5 230 that is so completely useless I disabled it.
Blitzvogel
Platinum Member
- Oct 17, 2010
- 2,012
- 23
- 81
I'm going to guess the IGP. M330 is only 320 Stream Processors with a 64 bit interface running to DDR3 VRAM. That's really stupid, but it might be an indicator of how many memory channels the APU is set up with. That's not a good sign at all :thumbsdown: It can't be cheaper to add even a crappy dGPU instead of using a higher end motherboard with no dGPU and dual channel support. Could be a higher config system on the way with much better graphics, that just isn't available yet, but that would preclude the idea of MXM based graphics.
What the **** is going on here?
Anyways, other than that, that Ideapad overall looks like a solid product when considering everything else (1080p screen!). Just would want to know what cTDP the processor is and if it's running dual channel. Would drop an SSD into it too of course.
On a side note - it has a VGA port.........oh lel
Last edited:
frozentundra123456
Lifer
- Aug 11, 2008
- 10,451
- 642
- 126
According to notebook check, the R5-M330 is 64 bit DDR3, 320 shaders, but is clocked fairly high, 1030 mhz. So it is probably similar the the igp of the FX, throttling ignored.
http://www.pcworld.com/article/3032...aptops-smoother-video-with-amds-freesync.html
if this was only a thing for the previous back to school and holiday shopping seasons...
if this was only a thing for the previous back to school and holiday shopping seasons...
Those prices are a little hard to swallow. $644 for a 14" one module A4-5150M??? I got a 13" Dell with Core i5 6200U for $499. Why would anyone get those HP laptops?
TRENDING THREADS
-
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 10K
-
Discussion Speculation: Zen 4 (EPYC 4 "Genoa", Ryzen 7000, etc.)- Started by Vattila
- Replies: 13K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 7K
Facebook
Twitter