With the rise of Deep Learning (DL), our world braces for AI in every edge device, creating an urgent need for edge-AI SoCs. This SoC hardware needs to support high throughput, reliable and secure AI ...processing at Ultra Low Power (ULP), with a very short time to market. With its strong legacy in edge solutions and open processing platforms, the EU is well-positioned to become a leader in this SoC market. However, this requires AI edge processing to become at least 100 times more energy-efficient, while offering sufficient flexibility and scalability to deal with AI as a fast-moving target. Since the design space of these complex SoCs is huge, advanced tooling is needed to make their design tractable. The CONVOLVE project (currently in Inital stage) addresses these roadblocks. It takes a holistic approach with innovations at all levels of the design hierarchy. Starting with an overview of SOTA DL processing support and our project methodology, this paper presents 8 important design choices largely impacting the energy efficiency and flexibility of DL hardware. Finding good solutions is key to making smart-edge computing a reality.
Abstract
The process of apoptosis is a critical component of normal immune system development and homeostasis, and in many cells this involves signaling through the c-Jun amino terminal kinase (JNK) ...pathway. In Jurkat T cells, Fas-induced JNK activity is dependent upon activation of the caspase cascades known to be central components of the apoptotic program. We show in Jurkat cell lines expressing a dominant negative PAK construct that PAK signaling is necessary for JNK activation in response to Fas receptor cross-linking. Inhibition of JNK activation induced by Fas does not impair cell death as assessed by DNA fragmentation. However, expression of the catalytically active C terminus of PAK2, which is generated through caspase action during Fas-mediated apoptosis, induces Jurkat cell apoptosis. We conclude that PAK activity resulting from caspase-mediated cleavage is a necessary component of JNK activation induced by Fas receptor signaling and that PAK2 can contribute to the induction of cell death.
Activation of p21-activated kinases (Paks) is achieved through binding of the GTPases Rac or Cdc42 to a conserved domain in the N-terminal regulatory region of Pak. Additional signaling components ...are also likely to be important in regulating Pak activation. Recently, a family of Pak-interacting guanine nucleotide exchange factors (Pix) have been identified and which are good candidates for regulating Pak activity. Using an active, truncated form of alphaPix (amino acids 155-545), we observe stimulation of Pak1 kinase activity when alphaPix155-545 is co-expressed with Cdc42 and wild-type Pak1 in COS-1 cells. This activation does not occur when we co-express a Pak1 mutant unable to bind alphaPix. The activation of wild-type Pak1 by alphaPix155-545 also requires that alphaPix155-545 retain functional exchange factor activity. However, the Pak1(H83,86L) mutant that does not bind Rac or Cdc42 is activated in the absence of GTPase by alphaPix155-545 and by a mutant of alphaPix155-545 that no longer has exchange factor activity. Pak1 activity stimulated in vitro using GTPgammaS-loaded Cdc42 was also enhanced by recombinant alphaPix155-545 in a binding-dependent manner. These data suggest that Pak activity can be modulated by physical interaction with alphaPix and that this specific effect involves both exchange factor-dependent and -independent mechanisms.
In a search for $\omega$ mesic states, the production of $\omega$-mesons in
coincidence with forward going protons has been studied in photon induced
reactions on $^{12}$C for incident photon ...energies of 1250 - 3100 MeV. The
$\pi^0 \gamma$ pairs from decays of bound or quasi-free $\omega$-mesons have
been measured with the CBELSA/TAPS detector system in coincidence with protons
registered in the MiniTAPS forward array. Structures in the total energy
distribution of the $\pi^0 \gamma$ pairs, which would indicate the population
and decay of bound $\omega~^{11}$B states, are not observed. The $\pi^0 \gamma$
cross section of 0.3 nb/MeV/sr observed in the bound state energy regime
between -100 and 0 MeV may be accounted for by yield leaking into the bound
state regime because of the large in-medium width of the $\omega$-meson. A
comparison of the measured total energy distribution with calculations suggests
the real part $V_0$ of the $\omega~^{11}$B potential to be small and only
weakly attractive with $V_0(\rho=\rho_0) = -15\pm$ 35(stat) $\pm$20(syst) MeV
in contrast to some theoretical predictions of attractive potentials with a
depth of 100 - 150 MeV.
The excitation function and momentum distribution of \(\eta^\prime\) mesons have been measured in photon induced reactions on \(^{12}{}\)C in the energy range of 1250-2600 MeV. The experiment was ...performed with tagged photon beams from the ELSA electron accelerator using the Crystal Barrel and TAPS detectors. The data are compared to model calculations to extract information on the sign and magnitude of the real part of the \(\eta^\prime\)-nucleus potential. Within the model, the comparison indicates an attractive potential of -(\(37 \pm 10(stat)\pm10(syst)\)) MeV depth at normal nuclear matter density. Since the modulus of this depth is larger than the modulus of the imaginary part of the \(\eta^\prime\)-nucleus potential of -(\(10\pm2.5\)) MeV, determined by transparency ratio measurements, a search for resolved \(\eta^\prime\)-bound states appears promising.
The
P
¯
ANDA fixed target experiment planned at FAIR will investigate fundamental questions of non-perturbative QCD. It makes use of a cooled antiproton beam (momentum: 1.5 to
15
GeV
/
c
) and will ...reach luminosities of up to
2
⋅
10
32
cm
−
2
s
−
1
, yielding a
p
¯
p
-annihilation rate of
2
⋅
10
7
s
−
1
. One option for the central tracker of
P
¯
ANDA is a cylindrical,
ungated,
continuously running TPC with GEM-based gas amplification stage.
Expression of the LAC9 gene from the yeast Kluyveromyces lactis in HepG2 human hepatoblastoma cells efficiently induced luciferase expression from reporter plasmids containing the four LAC9 binding ...sites from the K. lactis GAL1-GAL10 gene linked to a basal promoter. Induction was approximately 100fold and was dependent on the presence of the UAS sequence and an intact reading frame in the LAC9 gene. Additional cotransfection of constructs expressing the K. lactis GAL80 gene reduced luciferase activity by up to 98%. This inhibition was not affected by addition of 14mM galactose to the medium. No further yeast-specific factors appear necessary for efficient inhibition of LAC9 by GAL80, but additional gene products may be required for activation by galactose.