Intel to Discontinue First-Generation Thunderbolt 3 Controllers
by Anton Shilov on August 5, 2017 10:00 AM EST- Posted in
- Thunderbolt
- Intel
- Thunderbolt 3
Intel this week announced plans to discontinue its first generation Thunderbolt 3 controllers it launched in 2015, coming as a consequence of an industry transition from leaded to lead-free components. The company has had newer controllers in its fleet for quite a while, so it should not be a problem for PC makers to switch to them as they migrate to newer Intel platforms, such as Cannon Lake or Coffee Lake.
Intel on Thursday published plans to discontinue its DSL6340 and DSL6540 TB3 controllers it released in Q3 2015. Intel’s customers should place their orders on the chips by February 2, 2018, and the final shipments will be made by August 3, 2018. Replacing the DSL chips, Intel is advising its customers to instead use the JHL6340 and the JHL6540 controllers that were launched back in Q2 2016.
At first glance, there is no difference between Intel's DSL- and JHL-series Thunderbolt 3 controllers: both belong to the Alpine Ridge family and even their power consumption is the same: 1.7 - 2.2 W depending on port configuration. Intel has also confirmed that the DSL- and JHL-series TB3 controllers are similar in terms of features and functionality: 6340 supports one Thunderbolt 3 port, whereas 6540 supports two Thunderbolt 3 ports, every TB3 chip carries two DP 1.2 streams and so on.
| Intel's Thunderbolt 3 Controllers Codenamed Alpine Ridge | |||||||
| DSL6240 | DSL6340 | DSL6540 | JHL6340 | JHL6540 | |||
| Launch Date | Q2 2016 | Q3 2015 | Q2 2016 | ||||
| TDP | 1.2 W | 1.7 W | 2.2 W | 1.7 W | 2.2 W | ||
| Number of Ports | 1 | 2 | 1 | 2 | |||
| DisplayPort | 1.2 | ||||||
| Package Size | 10.7 × 10.7 mm | ||||||
| Recommended Price | $6.45 | $8 | $8.55 | $8 | $8.55 | ||
Meanwhile, there is a difference between how different families of Intel's Thunderbolt 3 controllers are made. The DSL-series controllers use a lead-containing solder alloy, whereas the JHL-series use a lead-free solder alloy based on tin, silver and copper (such alloys are called SAC - Sn, Ag, Cu). The European Union restricts the use of lead (as well as many other hazardous materials) because its fumes increase risk of lung and stomach cancer, along with the other known risks of lead exposure and heavy metals in general. So, as it appears, Intel is EOLing its first-gen Thunderbolt 3 controllers as part of the broader effort to phase out the use of lead in electronics products.
Related Reading:
- AKiTiO Displays Thunderbolt 3 to 10GBase-T Adapter
- LaCie Announces 2big Dock: 2-Bay TB3 DAS with Card Reader, USB-A and DisplayPort
- ASUS ZenBook 3 Deluxe UX490UA: Kaby Lake, 16 GB RAM, 1 TB SSD, TB3, 14" FHD, 2.4 lbs
- CES 2017: GIGABYTE’s Thunderbolt 3 to 8x USB 3 Dock
- LaCie Launches Rugged and d2 Thunderbolt 3 Storage Devices
- AKiTiO Introduces Node: Thunderbolt 3 eGFX Box for $299
Source: Intel
Facebook
Twitter
RSS
50 Comments
View All Comments
zdw - Saturday, August 5, 2017 - link
This is pretty boring news - the more interesting news is DisplayPort 1.4 support the next generation of Thunderbolt controllers that Intel is releasing with Coffee Lake CPU generation, which will allows higher resolution/higher framerate displays.repoman27 - Saturday, August 5, 2017 - link
Yes, that certainly will be interesting next year when there are CFL-S (that will still only support DP 1.2 because the graphics are the same gen as KBL) systems available with CNL PCHs (instead of the KBL-R PCHs that will ship later this year) paired with DP 1.4 capable Titan Ridge Thunderbolt 3 controllers that need to be connected to discrete GPUs so they can actually output a DP 1.4 signal. Then all we'll have to do is wait for the eDP 1.5 spec to be released, DP 1.4 / eDP 1.5 scalers / Tcons / panels to make it into production, and finally DP 1.4 displays to become available.CSMR - Sunday, August 6, 2017 - link
These are way too power hungry for widespread use. 1.7 W for single port. That is a significant portion of a TDP of a laptop. Intel can produce entire processors with graphics with a 4.5W TDP (Core M).Thunderbolt is trying to do far too much by combining PCIe and DisplayPort. These should be separate.
repoman27 - Sunday, August 6, 2017 - link
Thunderbolt is miraculous on power. That being said, an external I/O interface that is expected to provide two 20.625 Gbit/s full-duplex serial channels and 15 W of bus power per port is not exactly targeting the same platforms as a 4.5 W SoC. Compare the Thunderbolt 3 TDP to any other USB 3.1 xHCI (since it contains one of those) or 40 GbE controller. But if your definition of widespread use is more than hundreds of millions of devices, then yes.Why do you think the power required to maintain a 40 Gbit/s serial link would be any less if Thunderbolt only included PCIe protocol adapters and omitted the DisplayPort ones? This is like arguing that Ethernet is trying to do far too much by combining TCP and UDP. Packets are packets, and the DisplayPort silicon is completely shut down if you don't have a DP device connected. PCIe and DisplayPort are the fundamental protocols used by CPUs and GPUs, have similar lane rates, and are the two highest bandwidth I/O interfaces in common use on consumer PCs. It makes all the sense in the world to combine them to create a single external port solution. There is zero benefit in keeping them separate.
CSMR - Sunday, August 6, 2017 - link
Actually external pcie is most needed for mobile devices, rather than workstations which can use internal pcie. Yes I would expect that the complexity of multiplexing two different systems will increase power consumption. I cannot immediately find figures but would be surprised if a displayport connection to close to 1/10 of this power. Yes I would expect pcie to take more power than displayport at max throughput.Displayport is a standard that works very well. Hacking it together with pcie in a single connector imposes costs on users through hardware costs, cable confusion, power consumption, and greater room for unreliability. The problem that it solves for Apple (making devices with only one port) is Apple-specific and shouldn't be forced on PC users. External pcie should be additional to a couple of displayport ports, and so does not need to multiplex displayport.
repoman27 - Sunday, August 6, 2017 - link
Thunderbolt muxes PCIe and DisplayPort at the packet level, not the signal level. While the PCIe and DP connections from the Thunderbolt controller to the CPU / GPU / PCH inside the host might register on the power budget, they've got nothing on the PHYs for the external ports. Because it takes way more power to push 20 Gbit/s across a friction-fit connector and down a couple meters of cable than it does to move 8 Gbit/s along 10 cm of precisely laid out copper traces on a PCB. Whenever a Thunderbolt port is using the Thunderbolt signaling mode, the power required to maintain the link remains relatively constant, and it makes zero difference whether the packets being transported across that link are PCIe or DisplayPort.And don't worry, Thunderbolt is in no way being forced upon PC users. It will continue to be ignored by most OEMs as it is today because it costs too much and really was engineered primarily to meet Apple's design goals. Even if / when Intel integrates Thunderbolt into their chipsets, the PHY will still be a separate adder that will probably limit support to 2% of non-Apple devices. So you can rest easy knowing that proprietary dock connectors and dedicated HDMI ports will remain a feature of PCs for years to come.
mois - Saturday, July 28, 2018 - link
repoman27, do you know if the laptop's dual-port Thunderbolt 3 controller can process 32gbps, or just 22gbps, of sum aggregate PCIe data to&from the two TB3 ports at the same time?respond to roger1- in the forum PM
also, do you know if Akitio Node Pro (which has a daisy chain port and is not eGFX certified) and Razer Core Pro should have similar performance?
The Node Pro reviewer here
https://egpu.io/forums/thunderbolt-enclosures/unbo... https://egpu.io/forums/thunderbolt-enclosures/akit... insists that there should be some firmware difference that makes the Node Pro slower, however, I'd love the Node Pro for the practicality of having an extra TB3 port.
Please let me know what you think :)
HollyDOL - Monday, August 7, 2017 - link
It's good there is less lead used in things around, but out of those replacement materials tin isn't exactly the best thing under the sun either. I hope it isn't just hair of the dog because of contemporary popular green fashion.Wolfpup - Monday, August 7, 2017 - link
It's 2017 and we're only doing this NOW?!?Ugh, well, better late than never...
peevee - Wednesday, August 9, 2017 - link
"The European Union restricts the use of lead (as well as many other hazardous materials) because its fumes increase risk of lung and stomach cancer, along with the other known risks of lead exposure and heavy metals in general."What fumes? What exposure? The tiny little bit of lead is inside a chip, nobody is eating or huffing it, and each chip is soldered by machines.
EU bureaucracy is unbelievable...