Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
MP 7,1 Waiting for Mac Pro 7,1 A1991 (no more)
- Thread starter Mago
- Start date
-
- Tags
- macpro 2016
- Sort by reaction score
I guess it was't just me that noticed the different settings for 1080 and XFed 480s. The 480s had lower detail levels for sure.
All I'm saying is that I wouldn't easily call nMP targeting to musicians. An iMac with a consumer mobile-class gpu or a buffed MBP would be a more justifiable choice instead of a workstation with ws-class gpus (even more since they are also dual-only). Even a single D300 would be an overkill, and a Xeon cpu unnecessary. It's just not the ideal tool for the job, that's all.
By reading the logical assumptions here, it also seems that the 7,1 won't change direction on that matter (e.g. at least an option for a single-cpu config , so they can widen their target group a bit).
8GB version of RX480 at 229USD seems good value.
If it gets to 5.5TFlops it's great.
Speed seem to be at 1266MHz, according to TPU:
Default or OC'ed is anyone's guess.
Those waiting on 970/980 coming down in price with the launch of 1080/1070, there's a reason to hold on for a while longer.
I guess NVidia might have to lower prices at least of the 1070 now, although it's still on another league.
If it gets to 5.5TFlops it's great.
Speed seem to be at 1266MHz, according to TPU:
Default or OC'ed is anyone's guess.
Those waiting on 970/980 coming down in price with the launch of 1080/1070, there's a reason to hold on for a while longer.
I guess NVidia might have to lower prices at least of the 1070 now, although it's still on another league.
Define the job then.
When you work with a full orchestra of VIs you want, and need a lot of cpu power. Not to mention compression, EQs, various realtime effects, etc. Plus the realtime performance/playback of your DAW - plus video tracks(s) (I work in film music), and I could go on. An iMac is certainly NOT the ideal tool for the job! I couldn't do what I do each day on an iMac. If the 7,1 is announced I'll definitely look into it. The only reason I didn't get a 6,1 was the lack of PCI. Maybe I'll reconsider if the 7,1+extrernal PCI is a viable option.
Unless you're talking about the indie pop band that only needs a laptop, a guitar and a cheap mic when they record outside next to their westfalia.
Those 1266 might not be final, could be overclocked or AMD doesn't want to provide final specs just yet.
Judging by 10x0 clocks, if AMD can also take 4x0 higher around the same mark, the flagship model will be great.
I'm curious regarding the DP ratio.
Judging by 10x0 clocks, if AMD can also take 4x0 higher around the same mark, the flagship model will be great.
I'm curious regarding the DP ratio.
Some Xeon are specifically offered at server market other at WS market, the truth is users combine as they need you have desktops with server cpu and servers with desktop cpu, the fact is a 8 core high performance WS cpu zen is comming, as you surely read on my first post, it's more likely a Xeon Mac Pro this year (next year maybe the year of Zen Macs.
Until there is an independent benchmark on actual Zen silicone all these are just speculations.
!0GbE Thunderbolt networking is for daysi chained TB clients (utpo 6 computers using its 10GB MAC), on a Peer to Perr it can reach 20Gbps bi-di not using the Ethernet MAC. Hoever I'm in debt with you on the paper detailing this feature, it requires a bit deep search on detailed specs.
Yes, you miss the good part...
Even being a single system, this means a Several PCIe devices communicates efficiently each other and with the host.
This is a particulas case, but actually every CPU has an internal PCIe swtch inside doing oversuscription, what they do is to share buffer I/O bandwidth the External PCIe Switch what does is to expand this (adding latency as collateral, but this has become insignificant with latest PEX as 8xxx and 9xxx)
Worth you read the first page to educte yourself on PEX switches, while you (as ussually) are rigth on most your observations, there are few things you miss, and you need to account as system designer-wannabe (sorry if it may sound depreciative, actually I have great respect on you, as ever you gift us solid opinions hardly to rebut, no just logic fallacies as others).
http://www.anandtech.com/show/9060/the-asrock-x99-extreme11-review-eighteen-sata-ports-with-haswelle
As quoted on the PEX9xxx article, PCIe switches are designed to be transparent and most users isn't aware their existence, ASROCK and Gygabyte WORKSTATION Motherboards are proof of that, futher even those boards have been using to build monster gaming rigs, Supermicro has Several GPU Servers with 8 slots for 2 E5v3/v4 actually having PEX switches on it, HPC Clusters are among their favorite customers.
certainly the new PEX switches introduced several HOST-HOST fabric features good for HSA systems, not the case for single socket WS and I was obviating this.
Typically DMA transfers between peripherals are coordinated by the CPUs internal Switch and requires all the data going up/down the cpu while not touching DRAM buffers, previously PCIe switched didn't do anything about a DMA transfer required to share the CPU PCIe lines bandwidth, the newer PEX switched DMA transfers actually dont reach the cpu PCIe lines, are resolved at PEX, so its like having all PCIe slots concurrent (actually switched), on some situations its an improvement to the system native capabilities.
Most of the PEX9XXX innovations are aimed at multi-socket system, but two specific are major importance for single socket: much lower latency (in case 8xxx series it implies a performance penalization near to 1%, with PEX-9XXX it should be lower than on 8xxx.
In case latency still significant for GPU perfomace amid the DMA improvement, Apple could restrict the PEX switches only to TB3/NVMe where its less noticed, or TB3 alone if they use PCIe2 NVMe's (the latest is the fastest Mac possible, but implies using internal USB3 for Wireless/BT and using only the integrated 1GbT Lan, enforcing use Thunderbolt Etherner for those requiring faster links.
Last edited:
So you bought a Mac Pro just to encourage Apple's quest for coolness or something?
The target for the Mac Pro has always been the same and that won't change as long as there are Mac Pros.
Only in one sense, the target for Mac Pro is deemed for professionals. The landscape has changed. Software and hardware advances mean computers are used for things that were never envisioned back then.So you bought a Mac Pro just to encourage Apple's quest for coolness or something?
The target for the Mac Pro has always been the same and that won't change as long as there are Mac Pros.
The Mac Pro, is best for anyone that needs significant compute power from a GPUs and is optimised for FCPX. Right now they say that they aren’t for gaming, well guess what if the software is written such that games make use of parallel processing and run well on it then suddenly it’s a gaming computer.
I still maintain that Apple love to feel cool and suggest it’s for creatives. They may well be excellent for genome sequencing, are those guys in their labcoats what you’d call creatives?
Define the job then.
When you work with a full orchestra of VIs you want, and need a lot of cpu power. Not to mention compression, EQs, various realtime effects, etc. Plus the realtime performance/playback of your DAW - plus video tracks(s) (I work in film music), and I could go on. An iMac is certainly NOT the ideal tool for the job! I couldn't do what I do each day on an iMac. If the 7,1 is announced I'll definitely look into it. The only reason I didn't get a 6,1 was the lack of PCI. Maybe I'll reconsider if the 7,1+extrernal PCI is a viable option.
Unless you're talking about the indie pop band that only needs a laptop, a guitar and a cheap mic when they record outside next to their westfalia.
Yes, jobs may vary. But it's easy to understand that a machine that forcing you to get it with dual-only gpus (and there's not even a BTO option to reduce them) is narrowing its target group a lot, towards graphics professionals and a lot less towards musicians. That's plain common sense.
However, I'll take it that did you tried to do the same job with an iMac and failed, then ? And that was because the speed difference between nMP's Xeon cpu and iMac's latest i7 is that big ?
Aiden, whatsup? this is the PEX-switch for the TB2 controller on the nMP6,1, we where talking about the 8 PCIe2 lines from the PCH C602 on the nMP6,2 which has 4 wired to the SSD, 2 wired to the dual GbE controller and one Wired to the WiFi/BT module, leaving only one line PCIe2 free.
a nMP based on C612 not using a PEX to share the 8 free PCIe3 lines from the CPU, and enabling only 4 TB3 ports would require all that 8 PCIe2 lines from the PCH to drive the NVMe ssd(s), but given the C612 provides also USB3, it could wire unused USB3 ports to WiFI/BT
Last edited:
The posts that I replied to were discussing the MP6,1. Apple left four PCIe 2.0 lanes unused on the MP6,1.
The MP7,1 will be interesting, since each T-Bolt3 controller needs twice as much bandwidth.
I assume those are respectively the 4gb(199$) and 8gb(229$) versions.
I figure the nMP7,1 to have a slightly underclocked version of the 4GB model rebranded ad D310, and the 8GB version Rebranded as D510, introducing Vega as D710 later as the Darkent source commented, I would preffer to have an lower end 8GB Vega as D510, lets see what Apple thinks.
[doublepost=1464785903][/doublepost]
You're wrong Aiden, take your time to read the article from Arstechnica on the MP6,1 it details onlu 1 PCIe 2 line was left unused.
Beg to differ.
Only 3/4 of the bandwidth of the 8 PCIe 3.0 lanes going into the PEX are used - 25% of the bandwidth is left on the table.
ahh, that's another question, you talk anbout the 4 PCIe2 lines from the PEX Switch, sorry again those are used by the USB3 adapter since the C602 don't provide USB3.
Still sistem wide, the nMP 6,1 only left 1 PCIe2 unused.
Last edited:
The above diagram does not agree with what Anandtech found by mapping the PCIe complexes.ahh, that's another question, you talk anbout the 4 PCIe2 lines from the PEX Switch, sorry again those are used by the USB3 adapter since the C602 don't provide USB3.
Still sistem wide, the nMP 6,1 only left 1 PCIe2 unused.
In any event, there are four unused PCIe 2.0 lanes in the MP6,1. Either four from the PEX, or three from the PEX and one from the PCH. (Notice the unused PCH lane on your diagram.)
No I didn't try that on an iMac so I can't compare. Like I said I need PCI (for now), and the more cores I get the happier my project are, so I never really considered the iMac as an option.
[doublepost=1464788950][/doublepost]
Well, then at least they are all "pros", whatever that means.
I work in a certain field and I'm not going to pretend I know people's needs outside of my experience/knowledge.Since Apple has shifted its entire mac product with discrete GPUs to AMD it would make sense to discuss upcoming AMD product releases. There are a handful of people who consistently troll these threads and turn them into flame wars every time. I would encourage everyone to not engage with people exhibiting obvious trolling behavior.
I could also see a GPU lineup looking like:
D310: Polaris 11 XT, 4 GB VRAM, 7 TFLOPS total
D510: Polaris 10 cut, 8 GB VRAM, 11 TFLOPS total
D710: Polaris 10 XT, 16 GB VRAM, 12-13 TFLOPS total
There is a rumor on the front page that says the thunderbolt display is in short supply in Apple stores. Hopefully it means we are getting a 5k thunderbolt 3 display at WWDC! Something strange to me is several websites have reported on this but when they speculate on a 5k display they think its impossible because skylake doesn't support displayport 1.3. It seems pretty clear to me that a 5k display would be driven by 2 displayport 1.2 streams via a single thunderbolt 3 cable.
Agreed. I don't think Apple will announce a Mac Pro with a multi-month ship date. If they announce it at WWDC we are getting a mac pro with Polaris (or another currently shipping GPU).
Thats because we were getting a significant form factor change. You have to give developers a little time to prepare their workloads for dual GPUs and to use OpenCL. For just an update of the specs they will announce it within a couple weeks of the ship date.
8GB version of RX480 at 229USD seems good value.
If it gets to 5.5TFlops it's great.
Speed seem to be at 1266MHz, according to TPU:
Default or OC'ed is anyone's guess.
Those waiting on 970/980 coming down in price with the launch of 1080/1070, there's a reason to hold on for a while longer.
I guess NVidia might have to lower prices at least of the 1070 now, although it's still on another league.
AMD is going to kill it with this RX 480. $200 is a great price point and it will be an easy buy for a lot of people if it is only 80-90% as fast as the GTX 1070.
There is still a lot we don't know about Polaris 10 and the RX 480. I think its clear this is a cut down chip since they are pricing it so aggressively. Lisa Su also mentioned their will be a $300 Polaris part which would fit with a 8 GB full Polaris 10. It wouldn't surprise me if Apple got first dibs on the full chip and AMD waited 3-6 months to announce a RX 480X. We also don't know how capable it is at dual precision compute.
Macvidcards posted a picture that suggests that Anandtech was incorrect in their PCIe mapping and that Mago's diagram is right. The USB 3 controller is off the PCIe switch.
My bet is that the new mac pro will still need a PCIe switch and they will put both GPUs behind a single x16 PCIe lane. AMD already does this on their dual GPU cards and they need a lot of bandwidth due to the added lanes required from a faster SSD and each thunderbolt 3 controller. Either that or maybe AMD has a proprietary interconnect that Apple could use to directly attach the GPUs and then each would have a 8x connection to the CPU.
Macbook pro leaks show 4 USB-C connections and no standard USB-A connectors. If the mac pro follows suit than it will need all the extra PCIe bandwidth they can get if they do something like 10 USB-C connectors.
I believe the USB3 controller is using the 4 additional lanes out of the PEX, not as Anandtech says that it only takes 1 PCIe2 lane. that would be serious bad design choice.
Or better yet, report it. Check the forum rules here: http://guides.macrumors.com/Help:MacRumors_Rules and use the report button after each post. You'll see a few of the forum members have been suspended over the previous thread closing. So if things get uncivil, point it out to the mods.
The USB 3.0 controller is a Fresco Logic FL1100EX - which is a PCIe 2.0 x1 part.
https://www.ifixit.com/Teardown/Mac+Pro+Late+2013+Teardown/20778 (goto "step 20")
http://www.frescologic.com/products_show.php?ms=2&ps=10
![14060232441939631747[1].jpg](https://forums.macrumors.com/data/attachments/549/549007-3437b12d22e93eb8341a9077d5ab5cb8.jpg "14060232441939631747[1].jpg")
I mean, especially as we're talking about people here in certain fields that are not GPU-focused, I don't see an issue with offering a 4GB card as the entry level. It's still going to be a substantial step up from the current GPUs and for properly optimized tasks will scream along (just like how my old 7950 punches well above its weight still in FCPX compared to other programs.)
If you need more GPU power, and especially if they offered an ECC option, then you've got your built-in "good/better/best" incentive to upgrade.