Search

Predicting performance degradation of a GPU application at co-location on a spatial multitasking GPU without prior application knowledge is essential in public Clouds. Prior work mainly targets CPU co-location, and is inaccurate and/or inefficient for predicting performance of applications at co-location on spatial multitasking GPUs. Our investigation shows that hardware event statistics caused by co-located applications strongly correlate with their slowdowns. Based on this observation, we present Themis with a kernel slowdown model (Themis-KSM), which performs precise and efficient online application slowdown prediction without prior application knowledge. The kernel slowdown model is trained offline. When new applications co-run, Themis-KSM collects event statistics and predicts their slowdowns simultaneously. In addition, we also propose a two-stage slowdown prediction mechanism (Themis-TSP) for real-system GPUs without any hardware modification. Our evaluation shows that Themis has negligible runtime overhead, and both Themis-KSM and Themis-TSP can precisely predict application-level slowdown with prediction error smaller than 9.5% and 12.8%, respectively. Based on Themis, we also implement an SM allocation engine to rein in application slowdown at co-location. Case studies show that the engine successfully enforces fair sharing and QoS. • Themis predicts and reins in application-level slowdown on spatial multitasking GPUs. • The kernel slowdown model (KSM) collects hardware event statistics by per-SM counters and predicts slowdown with a pre-trained neural network. • The two-stage slowdown prediction model (TSP) predicts slowdown on real-system GPUs, by using a two-stage piecewise function to approximate kernels' SM-IPC curves. • The SM allocation engine reins in application slowdown to fulfill user requirements such as fairness or QoS. [ABSTRACT FROM AUTHOR]

The protein GSDMD is an important performer of pyroptosis and a universal substrate for the inflammatory caspase. However, the role and regulatory mechanism of GSDMD in Aspergillus fumigatus keratitis is remains unknown. Here we detected GSDMD protein in the cornea of normal and fungal-infected C57BL/6 mice. Human corneal epithelial cell (HCECs) were preincubated with a hydrochloride solution (IFNR inhibitor), ruxolitinib (JAK/STAT inhibitor), belnacasan (caspase-1 inhibitor) before infection with A. fumigatus conidia. Mice corneas were infected with Aspergillus fumigatus after pretreatment of GSDMD siRNA via subconjunctival injection. After, samples were harvested at specific time points and the expression of GSDMD and IL-1β was assessed by PCR, Western blot and immunofluorescence staining. Compared with the control group, we observed that the expression of GSDMD in fungal-infected mice cornea was significantly increased. After pretreatment with IFNR, JAK/STAT and caspase-1 inhibitors before fungal infection, the expression of GSDMD was significantly inhibited compared to the DMSO control in HCECs. Moreover, the GSDMD siRNA treatment have significantly weaken corneal inflammatory response, decreasing the proinflammatory factor IL-1β secretion and reducing neutrophils and macrophages recruitment in mice infected corneas. In summary, the data here provided evidences that GSDMD, an executor of pyroptosis, is involved in the early immune response of A. fumigatus keratitis. Additionally, the inhibition of GSDMD expression can affect the secretion of IL-1β and the recruitment of neutrophil and macrophages by blocking IFNR, JAK/STAT and caspase-1 signaling pathway. The protein GSDMD may emerge as a potential therapeutic target for A. fumigatus keratitis. • GSDMD, an executor of pyroptosis, is involved in A. fumigatus keratitis. • GSDMD affects IL-1β secretion and neutrophil recruitment. • The protein GSDMD may emerge as a potential therapeutic target for A. fumigatus keratitis. [ABSTRACT FROM AUTHOR]

The rapamycin insensitive companion of mTOR (Rictor) is an essential subunit of mTOR complex 2 (mTORC2), maintains the integrity of the complex and functions as regulator of Akt full activation. Rictor has been implicated to be involved in growth and progression of malignancies, however, little is known about its expression and prognostic role in gastric cancer in particular. Therefore, we investigated the relationship of Rictor expression with clinical outcomes, together with pAktSer473 and pS6, two downstream substrates of mTORC2 and mTORC1, in 396 gastric cancer tissue samples via immunohistochemistry. The results showed that 74.0% and 55.8% of tumors were Rictor and pAktSer473 positive staining, respectively, which correlated well with each other. Patients with positive expressions had poorer overall survival and relapse-free survival compared with those negative staining. Both Rictor and pAktSer473 expression were associated with lymph node metastasis, TNM stage, and WHO grading. Rictor was also correlated with tumor size, depth of invasion, and tumor thrombus, while pAktSer473 was also correlated with distant metastasis. In spite of 67.4% expression rate was presented in gastric cancer tissues, no significant association was observed between pS6Ser235/236, representing mTORC1 activity, and clinicopathological features or prognosis. These results suggest that mTORC2/Rictor/pAkt may play a more important role than mTORC1/pS6 in tumor progression, which could act as a prognostic biomarker or potential therapeutic target in gastric cancer. [ABSTRACT FROM AUTHOR]